The hidden grounds argument

Here is the argument:

1. (I) There are doxastic states (Ds) such that Ds are rational in virtue of being implicitly based on epistemic grounds.

2. (II) In paradigmatic cases of religious hinge commitments (RHs), RHs are implicitly based on epistemic grounds.

This is a parity argument. It is based on the observation that religious hinges are on par with rational doxastic states that are themselves implicitly based on epistemic grounds. The argument has the following form: ‘If x ₁ is F in virtue of x ₁ being G and x ₂ is G, then x ₂ is F in virtue of x ₂ being G’. What makes this schema valid is the explanatory generality of grounding. Grounding is explanatorily general in that particular grounding claims entail the truth of general grounding claims. Schematically,

Explanatory generality of grounding is widely accepted in metaphysical literature as intuitively plausible.Footnote ⁵ Accordingly, I take it for granted without further argument.

It should be noted that the very structure of the parity argument for the rational status of religious faith can be found in Newman.Footnote ⁶ My version of the argument focuses on the parity that concerns a rationality-grounding property (implicit basing). Against this background, I support (I) by (a) introducing the concept of implicit epistemic basing and (b) showing that there are instances of doxastic states that are rational in virtue of being implicitly based. Then, I support (II) by (c) introducing Newman’s view on the implicit grounds of religious faith and (d) arguing that the view is highly plausible.

Implicit epistemic basing

Implicit epistemic basing is a variety of epistemic basing. ‘Epistemic basing’ is the placeholder for the relation between doxastic state D and the epistemic grounds that make D justified.

What is constitutive of epistemic basing is the subject of heated debate. The primary options are some suitable form of causal relation (Moser Reference Moser1989), meta-belief about one’s evidence (Leite Reference Leite2008; Tolliver Reference Tolliver1982), or some combination of these factors (Korcz Reference Korcz2000; Ye Reference Ye, A and Bondy2019). More recently, it has been proposed that belief is rational when it is a manifestation of the epistemic virtue (Wedgwood Reference Wedgwood, Silva and Oliveira2022). The causal theories of epistemic basing are naturally combined with various forms of access externalismFootnote ⁷ about doxastic justification. However, doxastic theories of epistemic basing fit more naturally with access internalism about doxastic justification. Since I focus on the rational status of religious commitments in this paper, I assume that the rational dimension of epistemic justification necessarily involves a subjective, internal perspective on one’s epistemic situation. This suggests that doxastic theories are more promising from the perspective of the Hidden Grounds argument than from the causal and virtue-theoretic accounts.

The basic motivation for these doxastic theories of epistemic basing is that purely externalist theories give falsely positive results. Consider, for instance, causal basing of reliably formed epistemic states in arational or pre-rational agents (for instance, animals and infants). It is prima facie plausible that suitable non-deviant causal relations make those beliefs or desires reliably formed, even if we have strong intuitions that those beliefs or desires are not rationally held.

Doxastic theories, however, seem to overkill the problem faced by externalists. Doxastic theories predict that the subject with the doxastically justified belief that p has an occurrent or dispositional meta-belief about their evidential situation with respect to p. In general, standard doxastic theories of epistemic basing impose the following doxastic requirement on epistemic basing.

The paradigmatic cases of rational epistemic basing which require meta-belief are cases of ‘explicit processes of deliberation’ (Leite Reference Leite2004, 231) or reasoning. In such cases, the epistemic agent is at least disposed to form second-level beliefs about their epistemic situation with respect to the first-level proposition. In such second-level states, the subject explicitly appreciates the evidence he or she possesses.

However, I think that this doxastic requirement gives falsely negative results in many intuitive cases of rationally grounded beliefs. More specifically, doxastic theories are unable to explain the rational status of implicitly based rational beliefs. By ‘implicit basing’, I mean the cases in which we have a strong intuition that the beliefs in question are rational at least, in part, because the subject bases the beliefs on evidence, even though the subject does not have an occurrent or tacit meta-belief about their evidential situation.

Consider the case of ordinary perceptual justification. Let’s suppose that you form the belief that there is a pig in front of you in response to having the perceptual experience of seeing a pig in front of you. Intuitively, then – granted that all other necessary ingredients of the justificatory structure are in place (e.g. you have no undefeated defeaters, you trust your sense perceptions, etc.) – you hold that belief rationally even if you do not have a meta-belief about the evidential status of your experience.

This point is intuitively true for occurrent meta-belief. Perceptual justification – except in very special cases – is not a deliberation in which we reflectively investigate the epistemic status of our experience by forming occurrent beliefs about it. Rather, in paradigmatic cases of rational perceptual belief, we take perceptual experience at face value without having any occurrent meta-belief whatsoever. Appreciation of perceptual evidence, if present, is implicit.

For this reason, it is more natural to account for the doxastic requirement in terms of dispositions to have a meta-belief about evidence. For instance, in the perceptual case, the claim is that, if the perceptual belief that p is doxastically rational, then the subject is disposed to believe (or know) by reflection alone what perceptual evidence for p they have. Although the dispositionalist reading of the doxastic requirement has some plausibility, I will argue against it in this paper.

My main argument is that the dispositional doxastic requirement blurs an important conceptual difference between arational cognitive agents (let’s call them arational zombies), who are unable to have rationally grounded beliefs, and epistemic agents with rational beliefs, who are not disposed to reflect on their epistemic situation. As it happens, I am capable of reflecting on my evidential situation with respect to the contents of my perceptual beliefs. But imagine my hypothetical unreflective twin, who shares exactly the same experiences and beliefs as me as well as the same pre-reflective self-awareness, but who is unable to reflect – i.e. to formulate second-order beliefs – about his evidential and perceptual situation. If the dispositional doxastic requirement is true, then my unreflective twin and the unreflective arational zombie are in the same epistemic situation. I find this result implausible, however. Indeed, I think that it is much more plausible that my unreflective twin and I are in the same boat when it comes to the rationality of our perceptual beliefs.

To assuage those with access internalist inclinations, I would like to suggest that reflection and meta-belief are neither the only nor even the most rudimentary form of internal access to one’s evidence. While I am not able to provide a full argument for this view here, I will say the following: I take the core access internalist insight to be that one cannot believe something for epistemic reasons if such reasons do not figure into their epistemic perspective, at least implicitly. However, this figuring into one’s subjective perspective is not only having occurrent or dispositional second-order beliefs about one’s situation. There are other, non-doxastic or even non-propositional forms of self-awareness that help constitute one’s subjective perspective. For instance, Moser (Reference Moser1989) argues that epistemic basing involves non-propositional de re awareness of the evidential support without thinking that the given evidence supports the target belief. I find Moser’s (Reference Moser1989) view to be the most minimalist interpretation of the internalist insight – that having perceptual evidence essentially involves evidence figuring in one’s subjective perspective in this sense.

One might object that, because the contents of perceptual experience are too similar to the contents of beliefs that are based on experience, they cannot be called ‘evidence’ (see Austin Reference Austin1962, 115). I do not want to quarrel about the word ‘evidence’ here. If one prefers to restrict the concept of evidence to considerations that support p but are not as close to p as experiential appearances of p, then let it pass. All I insist on is that perceptual beliefs are rational in virtue of the subject basing them on grounds provided by perceptual experience (even if perceptual experience alone is insufficient to provide such grounds), independently of the subject’s deliberation or reflection. The concept of implicit epistemic basing putatively grasps this type of epistemic basing.

Implicitly based rational states

Having introduced the concept of implicit epistemic basing, let me now support (I) by drawing your attention to particular cases of implicitly based rational doxastic states. As I suggested earlier, the rational perceptual beliefs of my hypothetical unreflective twin are of this sort. Still, it would be helpful to find actual-world cases of implicitly based rational beliefs that might unanchor our concept of doxastic rationality from actual cases where belief is only based on explicitly available epistemic grounds.

Consider deeply entrenched background beliefs: understanding people that we have known for a long time, semantic knowledge of our mother tongue, and grasp of folk psychology and folk physics. Intuitively, such attitudes are epistemically rational. Are they rational in virtue of being based on epistemic grounds? One might object that such attitudes are not so based, despite taking them to be doxastically rational. As Ralph Wedgwood observed:

Wedgwood’s own reply to this question is that, since he has no recollection of the experience through or testimony by which he learned that his grandmother’s name was ‘Diana Hawkshaw’, his belief about his grandmother’s name is not based on evidence. Crucially, however, Wedgwood’s (Reference Wedgwood, Silva and Oliveira2022, Reference Wedgwood2023) view is that – despite this lack of evidential basing – the belief is rationally grounded because it is a manifestation of epistemic virtue. I think that Wedgwood’s considerations against an explanation of doxastic rationality regarding implicit basing trade on a tacit assumption – namely, that epistemic basing is constrained by a doxastic requirement.

The tacit assumption seems to be plausible because our grasp of the concept of epistemic rationality is anchored in cases where epistemic agents make the structure of their justification explicit by providing their epistemic reasons. For instance, many adult human beings, when asked for reasons for their empirical beliefs, cite the relevant perceptual experience. In the case of the name of Wedgewood’s grandmother, the epistemic ground is unavailable for reflection due to psychological reasons (for instance, the inability to recollect). I think, however, that even if this fact explains why the doxastic constraint on epistemic basing seems to be attractive, the appeal of this claim should be resisted. I offered support for my resistance to this claim in the previous section. So, even if Wedgwood’s question – ‘On what evidence is this belief now based?’ – does force us to revise the received doxastic view on epistemic grounding of rational beliefs, it can nevertheless be answered in a different way than Wedgwood suggests.

I contend that the better psychological description of the grounds for one’s belief about one’s grandmother’s name is that it is formed and sustained in response to considerations that support the truth of the belief (that one’s grandmother’s name is such and such). For instance, the belief might be based on direct testimonial evidence from one’s parents coupled with the understanding of verbal communication in the family, such that the hypothesis that ‘my grandmother’s name is such and such’ best explains this communication. This remains true, even if the subject is not able to recollect and reflectively access the grounds on which he or she formed the belief. Moreover, the grounds of deeply entrenched background beliefs are typically complex. The basing of such beliefs is a long and cumulative process occurring on the peripheries of one’s consciousness. These deeply entrenched background beliefs are gathered automatically, without explicit intention. In other words, epistemic agents typically base these beliefs in a manner that is mostly implicit.

Now, the key observation that motivates the implicit epistemic basing theory over Wedgwood’s alternative is that such implicit grounds figure into the subject’s perspective in an unreflective way. Having a different story and sequence of interactions with my family would, accordingly, affect my pre-reflective awareness of the facts that support my belief (that my grandmother’s name is such and such). It is true that I might not be able to recollect and reflectively access many of the relevant facts. Still, they determine my overall perspective. Moreover, the difference in my overall perspective counterfactually determines my disposition to believe that my grandmother’s name is such and such. If this is true, then implicit epistemic basing explains it better than the virtue-theoretic alternative, which seems to leave these facts unexplained.

What I find particularly interesting, in Wedgwood’s case, is that deeply entrenched background beliefs share much of their psychological profile with hinge commitments. Their deep entrenchment means that they constitute a bedrock of the belief system. Deeply entrenched background beliefs are firm and fixed. The subject is typically not able to provide the grounds on which they are based. Furthermore, such beliefs are relatively insensitive to rational revision. Revising one’s deeply entrenched belief (that one’s grandmother’s name is such and such) is not only strange – it requires a much more complex and powerful evidential basis than, for example, determining whether there is a pig in the garden on one’s perceptual experience.

This concludes my support for (I). Now, the second step of my argument is to support the parity premise that religious hinges are also implicitly based.

Newman’s extended evidentialism about religious faith

In this section, I interpret Newman’s view on the implicit grounds of religious faith as a form of extended evidentialism that involves an idea of implicit epistemic basing. Then, in the next section, I argue for the plausibility of this view.

Providing a comprehensive and concise reconstruction of Newman’s view on faith and reason is difficult. For a start, Newman seemed to care more about the richness of the detail of his psychological descriptions than the simplicity of his conceptual framework. For instance, in the preface to the third edition of The University Sermons (Newman Reference Newman2006, 9–10), he distinguishes between a ‘large and true’ sense and three ‘improper’ senses of the word ‘Reason’. In addition, he explains that his aim in The University Sermons was to contrast ‘Faith’ with ‘Reason’ in all three of these improper senses.Footnote ⁸ These distinctions are not entirely clear from the text of original The University Sermons, however.

Another source of potential confusion is that articulations of Newman’s views can change depending on the degree of polemical engagement on a given occasion. For instance, in one of his University Sermons, when arguing against rationalist views of his time, Newman boldly claimed that faith ‘is independent of processes of Reason’ (Newman Reference Newman2006, 129). Yet, a week later, after defining reason in the proper sense of the word as ‘the faculty of gaining knowledge upon grounds given’, he claimed that ‘if this be Reason, an act or process of Faith, simply considered, is certainly an exercise of Reason’ (Newman Reference Newman2006, 146).

As observed by Aquino (Reference Aquino2018) regarding University Sermons, Newman developed his own view by proposing a series of clarifications starting from initial formulations. Often, these initial formulations are influenced more by the popular understanding of the term to be defined (e.g. ‘reason’) or by rejection of the view that Newman contested (e.g. Lockean rationalist approach to faith). Over the course of his argument, Newman significantly mitigated the initial boldness and provided many important nuances to the proposed image. Based on this observation, I will follow this dialectic, reading Newman’s claims about the relation between faith and reason in light of the conclusions he eventually reached rather than isolated points he made at earlier stages of his argument.

In the above-mentioned preface to the third edition of University Sermons, Newman points out that ‘Reason’ – in the sense in which it is contrasted with ‘Faith’ in this work – is not reason in the proper sense of the word, but rather reason as popularly understood. Now, Newman’s own positive view on the nature of faith is that faith, when contrasted with reason in these three improper senses, ‘is implicit in its acts, adopts the method of verisimilitude, and starts from religious first principles’ (Newman Reference Newman2006, 10).

Let me briefly consider these three points. The first point of contrast between faith and reason, according to Newman, is that faith is ‘implicit in its acts’. This is the main theme of University Sermon 14, where Newman claims that ‘[f]aith cannot exist without grounds or without an object; but it does not follow that all who have faith should recognize, and be able to state what they believe, and why’ (Newman Reference Newman2006, 174–175).

In the Grammar of Assent Newman further elaborates this point, which is worth quoting at length:

The following point made by Newman is of great interest here. Simple or pre-reflective assent is identified by Newman as interpretative assent – i.e. assent that, when professed in the public act of faith, involves an argumentative process in which believers provide grounds for their assent ‘proportionate to their several capacities’. Crucially, the grounds provided in such an act of ‘interpretation’ only make explicit the previously implicit grounds of religious certitude on which such an act is based independently of it being interpreted and prior to it being publicly expressed. If this is the case, then such ‘interpretation’ does not establish a rational basing for such commitments but rather gives explicit articulation for such basing that is prior to and independent from being articulated. Even then, however, this articulation is not an infallible explication of actual implicit grounds, since people ‘may argue badly, but they reason well; that is, their professed grounds are no sufficient measures of their real ones’ (Newman Reference Newman2006, 149).

I propose a reading on which the pre-reflective nature of assent’s grounds is understood in terms of implicit basing, as discussed in the previous sections of this paper. The implicitness, read in these terms, does not mean that the disposition to reflect is essential for epistemic basing, but rather, that the reflective explication is always an ex post, more or less accurate articulation of the grounds of assent – grounds on which such assent is already based, independent of the believer’s ability to reflect on them.

The second point of contrast offered by Newman is that faith is based on ‘verisimilitudes’. Newman uses this term for epistemic grounds that fall short of being premises of ‘direct and definite proof’ (Newman Reference Newman2006, 134). Sometimes, Newman seems to restrict the term ‘evidence’ to that sort of direct and conclusive grounds (see e.g. Newman Reference Newman2006, 135; cf. 156). Such statements might encourage an anti-evidentialist reading, since, as observed by Aquino, ‘evidentialism of Newman’s day’ was that ‘one is rationally entitled to believe that p, if and only if one possesses full understanding of p and demonstrative proof of p’ (Aquino Reference Aquino2012, 19). However, evidence in contemporary epistemological literature is seldom understood as referring only to conclusive evidence that is offered by demonstrative proofs. Evidential support or fit, as it is understood today, also includes less than truth-guaranteeing forms of making a supported proposition more likely to be true. Therefore, it is natural to understand Newmanian grounds of faith (‘presumptions’) as regular evidence in the contemporary sense of the word.

The more substantial point made by Newman is related to the contrast between the probabilistic and inconclusive nature of grounds of faith and the maximal degree of certainty that is necessary for religious assent. This contrast results in some sort of lack of proportionality between faith and its epistemic grounds: ‘Faith … begins with its own previous knowledge and opinions, advances and decides upon antecedent probabilities, that is, on grounds which do not reach so far as to touch precisely the desired conclusion, though they tend towards it, and may come very near it’ (Newman Reference Newman2006, 156). If we expect rational agents to adjust their credences in p in strict proportion to the probability of p conditional on evidence being true, then this would suggest that faith cannot be rational.

In response, Newman insists that strict proportionality is an unrealistic requirement for doxastic rationality not only with respect to religious faith, but also concerning many cases of everyday assent: ‘In matters of daily life, we have no time for fastidious and perverse fancies about the minute chances of our being deceived’ (Newman Reference Newman1907, 192). Though, Newman concedes, absolute certainty in the absence of conclusive evidence is not irrational: ‘who of us would doubt, on seeing strong shadows on the ground, that the sun was shining out, though our face happened to be turned the other way? Here is faith without sight; but there is nothing against reason here, unless reason can be against itself’ (Newman Reference Newman1907, 192). The illative sense, on which he spent much time in the Grammar of Assent, is precisely the rational faculty that is involved in implicitly and pre-reflectively forming certitude in response to complex evidence that does not guarantee the truth of the conclusion.Footnote ⁹

Importantly, rejecting strict proportionality about evidential fit does not entail rejecting proportionality on the whole. For instance, assent might be proportionate to less than conclusive evidence if the evidential support reaches a certain threshold of cumulative power. Newman suggested that the complexity and richness of assent’s cumulative epistemic grounds in abductive reasoning (e.g. ‘ever-recurring experiences’ Newman Reference Newman2008, 67) allow meeting this proportionality requirement for ‘absolute adhesion of mind’, which in demonstrative cases is provided by valid proof (Newman Reference Newman2006, 247–248; see Aquino and Gage Reference Aquino, Gage, Aquino and Milburn2025, 37–40 for the role of cumulative cases in Newman’s epistemology).Footnote ¹⁰

In the Grammar of Assent, Newman emphasises the unconditional character of assent and faith in contrast to the acceptance of an inference, which is always conditional on the acceptance of the illative conclusion’s premises. This ‘unconditionality’ of religious faith seems to determine its phenomenology of certainty. It also seems to manifest in relative counterfactual independence between having reasons for p and assenting to p. After pointing out that assent is scarcely ‘ever given without some preliminary, which stands for a reason’, Newman stresses:

I will have more to say about how this ‘substantiveness’ of religious hinges fits their evidential basing in my responses to objections. For now, I would like to focus on the following. Newman clearly states that certitude is not the same as the unconditionality of assent. On the one hand, conclusions of demonstrative inferences are conditional despite being certain: ‘Inference is always inference; even if demonstrative, it is still conditional; it establishes an incontrovertible conclusion on the condition of incontrovertible premisses’ (Newman Reference Newman2008, 145). On the other hand, assent to the conclusion of the demonstrative inference, as all assent, is unconditional. Yet, this does not make Newman any more inclined to claim that such assent is not grounded in reason. As Newman warns his readers after contrasting assent with inference: ‘of course, I cannot be taken to mean that there is no legitimate or actual connexion between them [viz., inference and assent] … as if assent did not always imply grounds in reason, implicit, if not explicit, or could be rightly given without sufficient grounds’ (Newman Reference Newman2008, 145). All these points suggest that he issue of rational processes that give rise to assent and the issue of unconditionality of assent are, to a large extent, orthogonal in Newman’s work.Footnote ¹¹

The last dimension of contrast between faith and reason is related to different ‘principles’. In University Sermons, it is the difference between faith and reason in the third of the improper senses of the word – i.e. ‘secular Reason, or the “wisdom of the world”, that is, Reason exercising itself on secular principles in the subject-matter of religion and morals’ (Newman Reference Newman2006, 54, fn. 1). In Grammar, Newman refers to ‘first principles’ or ‘assumptions’ as the starting point of an inquiry into the given domain. ‘Religious first principles’ understood in this way likely also include assents that ground ‘dispositions for faith’, which make arguments for religious doctrine both ‘intellectually conclusive and practically persuasive’ (Newman Reference Newman1892, 74). In addition, Newman stresses that assent proceeding from religious first principles is not merely about an intellectual acceptance or ‘notional assent’ (Newman Reference Newman2008, ch. 4) – it necessarily requires acquaintance with the religious subject (Newman Reference Newman2006, 9). That said, we should remember that even real assent is for Newman an exercise of illative sense, which Newman takes to be an intellectual virtue that regulates the formation of certitude on less than demonstrative grounds (Newman Reference Newman2008, 281).

It is important to note that ‘reason’, with which Newman contrasts faith along these three dimensions, is only the improper, even if popular, sense of the word (Newman Reference Newman2006, 9–10). This strongly suggests that the resulting contrast between reason and faith is not absolute. It is rather a rejection of the over-restrictive kind of evidentialism, according to which epistemic grounds of religious faith are necessarily explicitly available, conclusive, and have the same nature as evidence in secular subject matters (see Newman Reference Newman2006, 10). If this interpretation is on the right track, then what Newman argues for is not anti-evidentialism tout court but extended evidentialism.Footnote ¹²

Quite recently, Duncan Pritchard clarified his earlier interpretation of Newman’s religious epistemology by proposing that Newman’s doctrine of illative sense is a form of anti-evidentialist virtue epistemology avant la lettre, which, even if not quasi-fideist itself, is compatible with quasi-fideism (Pritchard Reference Pritchard2025a). By ‘evidentialism’, Pritchard means here ‘something quite specific, albeit of a kind that would be familiar to philosophers of religion – viz., Lockean evidentialism, broadly conceived’ (Pritchard Reference Pritchard, Aquino and Milburn2025b, 11). However, as Pritchard admits (Pritchard Reference Pritchard, Aquino and Milburn2025b, 15) and as I noted above, this understanding of the term ‘anti-evidentialism’ leaves Newman’s position to be taken as a form of evidentialism in the less restrictive sense of the word. This Newmanian extended evidentialism sits well with contemporary evidentialism in appreciating non-conclusive and defeasible evidence. In other aspects, Newman offers important insights that potentially broaden the contemporary evidentialist framework (e.g. the idea of implicit grounds or rejection of strict proportionality as a requirement of doxastic rationality).

That said, Pritchard’s claim that Newman’s position is compatible with quasi-fideism seems to be more problematic. Newman’s epistemology seems to be compatible with a claim of hinge epistemology that every system of rational evaluation rests on fundamental arational hinge commitments which have no grounds in reason, implicit or explicit. Still, for quasi-fideism, we need something more – we need the claim that specifically religious hinges are arational in this way and that they share this arational profile with non-religious hinges. And this claim, as I argued above, is difficult to reconcile with Newman’s view on the necessity of implicit grounds for religious assent.

This gives us a general desideratum for interpreting Newman’s epistemology in virtue-theoretic terms. Interpreting Newman’s conception of illative sense in epistemic terms is a well-established approach (see Di Ceglie Reference Di Ceglie2023, ch. 1; Grimm Reference Grimm2001; Reference Grimm, Aquino and Milburn2025b; McNabb and DeVito Reference McNabb, DeVito, Aquino and Milburn2025). Granted, there are substantial differences between virtue-theoretic and evidentialist frameworks in general epistemology.Footnote ¹³ Nevertheless, the details of Newman’s epistemology, as reconstructed above, suggest that it represents a form of extended evidentialism which incorporates the theory of the epistemic virtue of illative sense, rather than posits opposition between epistemic virtue-theoretic and evidentialist approaches.Footnote ¹⁴

Implicit basing of religious hinges

Now, I want to argue that Newman’s theory of implicit grounds of religious faith is not just psychologically fine-grained but also – due to the fineness of its grain – highly plausible.

Let me start with the following observation. Since religious faith, according to Newman, is necessarily constituted by an act of simple assent, and since simple assent, due to its role in the structure of religious belief, might be identified with hinge commitment, Newmanian extended evidentialism might be naturally interpreted as a form of evidentialism about religious hinges.Footnote ¹⁵

To appreciate the plausibility of Newman’s proposal, let me compare religious hinges with the rational and psychological profile of deeply entrenched background beliefs. Like deeply entrenched background beliefs, religious hinges are deeply intertwined with other beliefs and intentions, and do not have reflectively accessible grounds. This intertwinement has two directions.

On the one hand, religious hinges qua hinges are individuated by constituting background assumptions for justification that the religious believer has for their ordinary religious beliefs and actions. For instance, the hinge commitment to the doctrine of Jesus’ resurrection is a part of Christians justification for the hope of redemption and eternal life. Or to give an example of a more ordinary propositional attitude being rationalised by a religious hinge, consider the fact that Catholics kneel down before what they believe to be the Blessed Sacrament, and this is rationalised by their hinge commitment to the doctrine of real presence of Christ in the Eucharist.

On the other hand, religious hinges are supported by testimony and participation in religious practice that provides complex, abductive support for their truth. For instance, being informed by a reliable source that Jesus has resurrected will rationally raise one’s confidence in the proposition that Christ has resurrected. Other sources of support for religious hinges are religious experiences and sense-making explanations of important phenomena based on religious cornerstones (see Alston Reference Alston1991, ch. 8 for an overview of the plausible candidates for the sources of evidence for religious beliefs).

Due to its complexity and mostly tacit form of gathering, the evidential basis for religious hinges is typically not fully available for reflection nor testimony. However, this lack of availability also exists in the case of many deeply entrenched background beliefs. For instance, Newman claims that assent to the proposition ‘that there are things existing external to ourselves’ is based, at least in rational human beings, on testimony of ‘ever-recurring experiences’ (Newman Reference Newman2008, 67). Note that the proposition ‘there is an external world’ is among the chief examples of hinge propositions (see Wittgenstein Reference Wittgenstein1969). Yet, when asked for evidence that there are things that exist for us, most of us will be perplexed, since no ready evidence for this proposition is reflectively available. Rather, the evidence is the totality of our experiences that supports this proposition.

This makes the acceptance of religious hinges implicitly, rather than explicitly, based on epistemic grounds, even in the case of otherwise reflective subjects. For instance, even in highly reflective believers who can reflect on their epistemic conduct, the assent to religious hinges is the result of a long, complicated, and primarily implicit process. Moreover, that process might be ex post rationally reconstructed. Importantly, one’s reflective abilities are not necessarily engaged in the very epistemic basing of religious hinges. Further, when employed, such reflective abilities can only, in a very general sense, ex post reconstruct the epistemic grounds of religious commitment. Rational autobiographical reconstruction of the religious conversion of highly reflective believers like St Augustine (2008), John Henry Newman (Reference Newman1963), C. S. Lewis (Reference Lewis1955), Peter Geach (Reference Geach and Lewis1991, 7), Michael Dummett (Reference Dummett, Auxier and Hahn2007, 4–5), and Edward Evan Evans-Pritchard (Reference Evans-Pritchard1973) supports this view.

If this is how epistemic basing of religious hinges works in reflective subjects, then we might claim that the epistemic basing of religious hinges indeed fits the Newmanian theory of interpretative assent. If what their reflection uncovers is sufficient to make hinges rational prior to an act of reflection, then hinges are implicitly based on epistemic grounds.

That kind of rational support is not necessarily conclusive. There is no guarantee that the cornerstones of the religious belief system are true given the evidence on which they are based. Still, even if implicit grounds do not entail the truth of the supported proposition, the subject is certain that the proposition is true – and rationally so. The same train of thought applies to deeply entrenched rational background beliefs. Quite often, such beliefs concern highly contingent facts which are not themselves necessarily true, even given the evidence that the subject has. Nevertheless, deeply entrenched background beliefs are formed in response to evidence that is gathered in the long, implicit, and cumulative process and they involve rational ‘absolute adhesion of mind’ to their contents.

This concludes my argument for the Hidden Grounds thesis. Let me now turn to the objections against the proposed view.

Objections and replies

Objection. The key insight behind the idea of hinge commitments is that rational support is constitutively local in the sense that necessarily not all propositions constituting the structure of subject’s justification are in fact justified (Pritchard Reference Pritchard2016, Reference Pritchard2021). More precisely:

Hinges are not then ‘in the market for rationally grounded knowledge’ (Pritchard Reference Pritchard2016, 89–90). The Hidden Grounds thesis misses this a priori point and introduces an internally incoherent conception of rationally grounded religious hinges.

Reply. The locality thesis is not uncontroversial. Simion et al (Reference Simion, Schnurr and Gordon2021) provide examples of possible cases in which hinge commitments are rationally revised in the face of counter-evidence. Neta (Reference Neta2019) provides an a priori argument for the claim that hinges are evidentially supported by the fact that one has rational support for empirical propositions. These considerations point to the fact that the locality thesis is problematic.

I think that the hinge/ordinary belief structure might be preserved without the locality thesis. In the case of religious hinges, I consider it to be a psychological fact that it is, for religious believers, that they lose or gain their religious faith at least partly because of epistemic reasons. Locality offers only an error theory for that intuition. I grant, however, that religious hinges – just like all hinges and deeply entrenched background beliefs – are relatively stable and less sensitive to counter-evidence than ordinary beliefs. Ideally, it follows that we should be able to satisfactorily explain the stability and evidence insensitivity of religious hinges without violating believers’ intuitions about the epistemic grounds of their commitments.

I believe that the weak locality thesis, together with the Hidden Grounds thesis, offers such an explanation. According to the weak locality thesis,

The thesis implies that reflectively available rational support is local, relative to the normal context of epistemic evaluation, but not to the hinge proposition itself. The concept of the normal context of evaluation functions here in a manner that is similar to Kuhn’s (Reference Kuhn1970, ch. 3) concept of normal science: in normal contexts, the hinges/paradigms are taken to be true without reflectively available evidence. Rather, they are firm and fixed hinges of the epistemic basing of normal belief/science. Still, hinges have the phenomenology of rational responses to evidence that one has. The point is that this phenomenology does not entail the ability to make the evidence explicit on reflection. However, it does not mean that hinge commitments cannot be evaluated in terms of epistemic basing and rationality.

For instance, in ‘abnormal’ contexts of crisis of belief, believers might rationally revise their hinges in the face of counter-evidence. Another example is when a believer makes reflective ascent and evaluates their own state in terms of rationality. Importantly, such ascent is a different, reflective enterprise than basing one’s hinges on epistemic grounds.

The weak locality thesis explains – without proposing an error theory – why it seems, from the perspective of the believer, that hinges are rationally maintained and revised even if the subject is not able to reflectively specify the epistemic grounds on which hinges are based.

Objection. Hidden Grounds thesis misrepresents the psychological and epistemic profile of religious commitments. As Duncan Pritchard observed,

Reply. In reply, I contend that the pull of this objection can be mitigated if we carefully examine what guides our intuitions about rationality in particular cases of rational belief revision.

The intuition is that ‘there would something seriously amiss with someone who professed to a faith in God, but who was nonetheless willing to abandon this commitment once faced with counter-evidence that she is unable to rationally dismiss’. Now, if by ‘rationally dismiss’ we mean ‘dismiss on reflectively available grounds’, the intuition is plausible but does not affect the Hidden Grounds thesis. If by ‘rationally dismiss’ we mean ‘dismiss on reflectively available (implicit) grounds’, its plausibility is at least partially undermined by the theory of implicit epistemic basing.

Consider the revision of deeply entrenched convictions that are implicitly based on complex grounds. For instance, naïve realism about the mind-independent existence of perceived objects seems to be a hinge commitment for virtually all adult human beings. Findings from developmental psychology (see Peters Reference Peters2022) suggest that children typically acquire the cognitive abilities necessary to operate with this distinction between perceived appearance and mind-independent reality at the age of four. Rational basing, if any, of such a hinge is mostly implicit and supports naïve realism with highly complex epistemic grounds. As Newman observed with respect to this case:

Now, consider epistemic agent A, who simply and unreflectively assents to naïve realism about perceived objects. A is confronted with an idealist, Berkeleyan argument against the mind-independent existence of perceived objects. Assume that such an argument is fallacious. Crucially, due to the simple character of assent and the general paucity of his reflective abilities, A is not able to find the fallacy in and thus defeat Berkeley’s argument. The theory of implicit epistemic basing predicts that A would be rational in sustaining his assent to naïve realism. Note that this prediction is fully compatible with the claim that A is not able to defeat the Berkeleyan argument by reflectively available grounds.

If the Berkeleyan argument is not fallacious, and the implicit grounds of A’s commitment to naïve realism are unable to defeat the argument, it is not rational for A to steadfastly adhere to naïve realism.

Given the truth of the parity claim (II), the prediction for religious hinges is the same. The Hidden Grounds argument is not ‘epistemic through-and-through’ in the sense that religious hinges are on par with ordinary beliefs in terms of their evidence sensitivity. The Hidden Grounds thesis predicts that rational revision of hinges is more difficult than that of ordinary beliefs due to impaired reflective command over their grounds. The grounds of ordinary beliefs are, at least in paradigmatic cases, reflectively available and less complex in terms of epistemic control. This facilitates rational management of beliefs based on such evidence. The rational management of religious and non-religious hinges as well as deeply entrenched background beliefs is much harder because it is difficult to assess and evaluate their grounds. However, ex hypothesi of implicit epistemic basing, this does not affect the rationality of these states.

Objection. Implicit grounds of religious hinges are not, properly speaking, epistemic. As Wittgenstein put it:

Now, the objection raised by some Wittgensteinians (even if not Wittgenstein himself) is that such ‘experiences, thoughts’ might constitute some reasons to revise religious hinges, but, importantly, ‘these reasons are not “evidence” in the sense that they epistemically ground the belief such that any rational agent would be forced to adopt it’ (Schönbaumfeld Reference Schönbaumfeld2023, 35).

Reply. Denying the epistemic character of the grounds of religious hinges might be either contrastive or absolutist. It is contrastive when its proponent argues against evidential support for hinge commitments by contrasting them with uncontroversially epistemic kinds of support. For instance, religious hinge commitments are often contrasted with ‘ordinary’ empirical beliefs that are uncontroversially based on perceptual evidence. The denial might be absolutist when the pragmatist argues that the grounds of religious hinge commitments could be explained away in non-epistemic terms, like the grounds of all hinge commitments.

I find it striking that some Wittgensteinian fideists in philosophy of religion often base their position on contrastive arguments. For instance, it is common to contrast religious commitments with empirical beliefs (see e.g. Schönbaumfeld Reference Schönbaumfeld2023, ch. 3). However, the deep structure of the Wittgensteinian critique of evidentialism in philosophy of religion is absolutist in spirit. This incoherence makes the fideist position somewhat unstable. For one thing, the contrastivist’s strategy is based on empiricist assumptions that have been abandoned in general epistemology. For some reason or another, such assumptions seem to find a nice refuge in philosophy of religion.

My main line of response lies at a different place, however. Below, I respond to the stronger absolutist argument. If my argument is sound, and granted that religious hinges are on par with non-religious hinges when it comes to their epistemic basing (II), then the contrastive argument is blocked as well. To fix terminology, let us agree that when an attitude is epistemically grounded, it is ‘grounded’, and when it is non-epistemically grounded in the quasi-Wittgensteinian sense of being forced by life onto us, it is ‘forced’.

The absolutist argument from forcing against the epistemic character of the basing of hinge commitments should be able to explain all relevant aspects of having hinge commitments in non-epistemic terms. I contend, however, that following aspects of hinge commitments escape explanation in terms of forcing.

Hinge commitments are assertoric attitudes towards propositions or more complex, holistic contents that the subject accepts as true. Accordingly, hinge commitments have truth conditions. Next, there are several processes and events, some of which are essential for having such commitments, that involve systematic counterfactual dependencies – probabilistic, explanatory, necessary – between the truth conditions of hinge propositions and the conditions of which the subject is aware. These counterfactual dependences are sufficient for the latter conditions to be the subject’s epistemic support for or against hinge propositions (propositional justification for hinge commitments). Next, the subject’s acceptance of a hinge proposition (hinge commitment) is best explained by the involvement of the subject’s awareness of these conditions. In other words, the rational support for the hinge proposition that the subject has (rather than the mere fact of being aware of it) is a causal factor that makes the subject accept the hinge propositions. As is clear from testimonies about religious conversion as well as loss of faith, it is common for subjects to report the phenomenology of rational belief revision: they cite reasons that bear on the truth of religious hinges as reasons for their change in belief system.

It is true that converts and apostates often describe the change in their views as being ‘forced’ by the general course of experience and thought, which is deeply intertwined with their other beliefs and intuitions. However, the theory of epistemic basing implies that it would be an overreaction to understand this in terms of non-epistemic forcing. As we have seen, deep entrenchment in the form of life and the resulting relative inaccessibility to reflection and lack of reflective command on one’s reasons for religious belief is anticipated by the Hidden Grounds thesis. In that sense, my thesis appears equally capable of accommodating this data as the explanation that appeals to non-epistemic forcing. At the same time, the Hidden Grounds thesis fits the data about counterfactual and probabilistic dependence that constitutes propositional justification for hinge change and the phenomenology of rational hinge revision. Non-epistemic forcing, at best, seems to offer an error theory for these phenomena. Overall, the Hidden Grounds thesis offers a more powerful and less revisionary explanation of religious hinge revision than the anti-epistemic theory of forcing. I think this is good reason to at least mitigate the force of the current objection, if not uphold the Hidden Grounds thesis over the theory of non-epistemic forcing.