Perceiving God like an angel

Abstract

Mystical experiences are often regarded as potential sources of epistemic justification for religious beliefs. However, the ‘disanalogy objection’ maintains that, in contrast to sense perceptions, mystical experiences lack social verifiability and are thus merely subjective states that cannot substantiate objective truths. This article explores a novel externalist response that involves the concept of angels. As spiritual beings, angels can directly perceive God and verify these perceptions in their celestial community. Thus, the ‘direct perception of God’ is not inherently incapable of social verification. While invoking angels might appear contentious, it coheres with the externalist approach of conceptualising cognitive states under hypothetical settings. Despite the differences between humans and angels and their lack of interaction for verification purposes, our approach remains valid because mystics not only exemplify the same general type of ‘direct perception of God’ as angels but can also be preliminary members of a wider celestial community.

Introduction: perceptions of God and the disanalogy objection

Mystical experiences have long been regarded as a source of justification for religious beliefs. When a mystic feels the presence of God, she is prima facie justified to believe in God on the basis of this experience (Brown 2015; Griffioen 2021; Hick 1989; Netland 2022). However, this proposal faces the well-known ‘disanalogy objection’, which argues that mystical experiences, unlike sense perceptions, are not socially verifiable and thus lack cognitive value. Sense perceptions can often be socially corroborated. For instance, the claim of seeing a rose in a garden can be verified by examining the perceiver’s optical-neurological system, the light conditions, and so on. No similar method exists for verifying mystical experiences of God. Although religious traditions have assessed mystical experiences through moral outcomes and doctrinal coherence (e.g. Iqbal 2013; James 2002; Wainwright 1988; Webb 2020), these criteria do not conclusively establish the authenticity of such experiences. Even if a mystic is generous and humble, her experience of God might still be merely a mental construct. Martin (1959, 67) thus complains: ‘There are no tests agreed upon to establish genuine experience of God and distinguish it decisively from the nongenuine’. The disanalogy objection posits that mystical experiences, lacking social verifiability, are mere ‘subjective’ states that cannot substantiate objective truths (e.g. Gale 1994).

Internalists have a direct response to this objection: on phenomenal conservatism, the experience whereby p seems true prima facie justifies p (Huemer 2011; Swinburne 2018); thus, the experience of God’s presence prima facie justifies beliefs in God irrespective of social verifiability (Kwan 2009, 535–536). Externalists, who regard sense perception as the paradigmatic source of justification, face greater challenges. In particular, Alston suggests that sense perception offers justifications as a ‘doxastic practice’, that is to say, a socially established belief-forming process subject to overriders, not proven unreliable, and so on. If mystical experiences critically differ from sense perceptions, their status as justification-apt doxastic practices will be suspect. Alston’s solution is to regard verifiability as inessential: ‘there is no reason to suppose it appropriate to require the same checks and tests for [mystical perceptions] as for sense-perceptual reports … Here we have what is perhaps our most glaring example of epistemic imperialism’ (1991, 216). Unfortunately, this move is problematic. While some disanalogies are irrelevant – such as whether mystical experiences are widely distributed as sense perceptions – verifiability seems essential to objective truths.¹

This article presents a novel externalist solution to the disanalogy objection by incorporating the notion of angel. As spiritual beings with superior knowledge of God, angels are uniquely positioned to verify perceptions of God: when an angel perceives God, other angels can verify this perception. Thus, the ‘direct perception of God’ is verifiable in the angelic community. This conclusion extends to human mystical experiences as well. Because angels possess greater insights into God’s will and are more knowledgeable about when God is revealed, they are better equipped to evaluate the veracity of human mystical experience of God. Contrary to the disanalogy objection, mystical experiences can indeed be verified. They are verified by angels. Admittedly, introducing angels may raise several concerns. First, it appears illegitimate to employ such disputable entities without evidence of their existence. Second, given the fundamental differences between angels and humans, it is suspect whether angelic perceptions are relevant to humans. Third, since we do not usually interact with angels, the appeal to angels cannot enhance the social verifiability of human mystical experiences. After outlining the disanalogy objection and the basic idea of our solution, we develop our view in response to these objections. We address the first issue by clarifying our externalist stance. Instead of resting our case on evidence about angels, we defend only a conditional claim: if angels exist, mystical experiences are verifiable. This proposal aligns with externalist approaches to the essence of cognitive states and effectively counters the disanalogy attack on the epistemic potential of mystical experience. To tackle the second issue, we argue that angels are relevant to mystics because both share the same general type of ‘direct perception of God’. The distinctions between humans and angels do not threaten this verdict. Regarding the third objection, it will be shown that mystics can be included in the angelic community as preliminary members. Therefore, human mystical experiences may not only be type-identical to angelic perceptions of God but also occur within the same extended celestial community in which they are socially verifiable by angels. In Augustine’s (2008, 191; Ep. 147. 22) words, we perceive God when we perceive ‘like an angel’.

A brief presentation of the disanalogy objection

To lay the groundwork for our angelic solution, we first clarify the basic concepts related to the disanalogy objection. Throughout this article, ‘mystical experience’ and ‘mystical perception’ are used interchangeably, and ‘mystic’ refers to all human perceivers of God, encompassing both contemplative mystics and everyday believers.

Sense perception is often considered the paradigmatic source of epistemic justification. Based on the observation that sense perceptions are frequently checked by ourselves and by others – explicitly or implicitly – one might suggest that a perceptual experience offers justification only when it has been cross-checked.² Obviously, this proposal is overly strong. While actual cross-checking might be necessary to establish the ultima facie justificatory value of an experience, it is not essential for initial, prima facie justifications. The disanalogy objection, disqualifying mystical experiences as incapable of cognitive value, is asserting that these experiences cannot offer even prima facie justifications because they are impossible to verify. Hence, the objection hinges not on the actual presence but on the possibility of cross-checking. For an experience with content p to prima facie justify p, it must be possible to properly cross-check this experience. More specifically, it must be possible to check not only the truth of p but also the cognitively adequate link – be it causal or modal – that the experience holds in relation to the truth of p without Gettier-styled lucks.

To be sure, it is relatively straightforward to falsify mystical experiences. If a Christian experiences what appears to be an encounter with God immediately after taking LSD, or if the experience results in arrogance rather than spiritual humility, we can reasonably conclude that the experience is not genuine. Even if God exists and is somehow the cause of this experience, the use of LSD and the absence of intended ‘fruits of the spirit’ (e.g., Alston 1991, 276) provide sufficient grounds for rationally regarding the experience as non-veridical.³ Hence, the real challenge posed by the disanalogy objection lies in verifying mystical experiences. When a mystic reports a direct experience of God, what procedures can reliably determine the genuineness of this experience? Acknowledging God’s existence is not a solution. Because God is the cause of all things, His mere existence does not distinguish mystical experiences that are genuinely caused by Him from those that are not. Furthermore, moral standards are insufficient for identifying genuine mystical experiences. Even if a mystical experience leads to moral improvement – such as a higher level of humility and generosity – it may still be a mere mental construct. This is clearly not the case for ordinary sense perceptions. When a person sees a rose in a garden, her perception is properly verified once we confirm that the rose is indeed present, that lighting conditions are normal, and that the person’s optical system is functioning properly. Such procedures allow us to conclude the veracity of the person’s perception unless we unduly embrace the sceptical idea that we are victims of a Cartesian demon or brains in vats. Thus, mystical experience and sense perception appear to differ regarding the following condition about ‘sufficient verifiability’ (SV):

SV: For an experience of content p to prima facie justify p, there must be sufficient evidence for the truth of p and the cognitive adequacy of this experience.

Alston acknowledges that SV applies to sense perception but concedes that mystical experiences do not meet equally stringent verification conditions. He notes, there are ‘no clear-cut conditions such that we are prepared to admit that God exists and is perceived by me if and only if a person who satisfies those conditions perceives God whenever God is present to him’ (1991, 214). This remark implies a related thesis about sufficient conditions for perception (SCP)

SCP: For an experience of content p to prima facie justify p, there must be sufficient conditions under which any agent will perceive p.

SCP, closely related to verification, clearly applies to sense perceptions. When I perceive a rose in a garden, others will also perceive it if they are present and looking in the right direction under optimal light conditions. Some epistemologists reject SCP as irrelevant to mystical experiences because whether God is perceived might eventually depend on whether He chooses to be seen.⁴ Still, SCP can apply to divine perceptions if we include God’s intention as a condition for perceiving Him. We may not know when God intends self-disclosure, but our ignorance does not invalidate divine intention as a condition for perceiving God. A potential worry is that God’s intention will pre-empt other conditions. The intention of the omnipotent God to reveal Himself necessarily entails that He will be perceived and thereby renders other conditions redundant. Yet, perceptual conditions need not be independent from one another. God’s intention might align with the perceiver’s spirituality, such that only the ‘pure in heart’ shall see God (Matthew 5:8). Although purity of heart may not be indispensable for perceiving God, it remains a reliable pathway to genuine divine encounters.

In this article, we presume both SV and SCP as plausible theses about justification. We do not debate which is more fundamental. Some epistemologists prioritise SCP as more fundamental, suggesting that mystical experiences fail SV because they fail SCP. Nevertheless, SCP should not be seen as a detailed account of verification: the process of verification can check the reliability of the relevant cognitive faculties without involving re-perceiving the object itself.

Verification, it should be noted, is inherently ‘social’. When an objective truth is perceived, the ‘evidence’ for verifying this perception, as in SV, should be available to others. As Gale (1994, 871) argues, it is a ‘conceptual requirement that the object of a veridical sense perception be perceivable by different perceivers’. This does not mean that all genuine perceptions can be practically verified by others, of course, for we can perceive truths when no one else is present to witness them (e.g., Mavrodes 1970, 79). Verifiability concerns the possible existence of supporting evidence, not its practical accessibility. This leads to a thesis about verifiability in possible societies (PS):

SV-PS: For an experience of content p to prima facie justify p, there must be sufficient evidence by which some other possible agents could establish the truth of p and the cognitive adequacy of this experience.

A stronger condition appeals not to other possible agents but to others in one’s actual society (AS).

SV-AS: For an experience of content p to prima facie justify p, there must be sufficient evidence by which some other agents of the experiencer’s own actual society could establish the truth of p and the cognitive adequacy of this experience.

SV-AS, as robust as it is, is endorsed by Alston (1991, 216) when he suggests that perceptual experiences with justificatory power must be open to scrutiny ‘by other members of the community’. He concedes that mystics are unlikely to gain justification from their experiences if they are ‘going it alone’ without belonging to any actual religious community (1991, 200). Let us call ‘P-social verifiability’ the possibility of verification by other possible agents and ‘A-social verifiability’ the possibility of verification by other members of one’s own society. Because other members of one’s actual society are also possible agents, A-social verifiability entails P-social verifiability.

For the sake of argument, we do not question SCP, SV-PS, or SV-AS. Rather, we use them to reconstruct the disanalogy objection as follows: mystical experience must satisfy the conditions in SCP, SV-PS, or SV-AS to offer justification; since it satisfies none, it lacks cognitive value.

The social verifiability of angelic perceptions of God

From an exclusively human perspective, we concede that mystical experiences indeed fail the conditions in SCP, SV-PS, and SV-AS. However, this does not mean that mystical experiences are inherently unverifiable. Rather, their unverifiability could be contingent on human limitations, and the conclusion might differ if our perspective is broadened beyond humans. In Christian theology, angels are depicted as directly perceiving God without the mediation of sensations or intellects. Matthew 18:10 states that angels ‘always behold the face’ of God, and Tobit 12:15 describes Raphael as standing ‘in the glorious presence of the Lord’. Angelic perceptions of God are commonly referred to as beatific vision, which Pace (1907) defined as ‘immediate knowledge of God which the angelic spirits and the souls of the just enjoy in Heaven’. Notwithstanding the debates about the scope of beatific vision, it is generally agreed that ‘direct perception of God’, broadly construed, applies to mystics, redeemed souls in Heaven, and angels. Albeit finite, angels have impeccable knowledge of God. Isaiah 6:1–8 portrays seraphim, the highest angels in the Pseudo-Dionysius system, as intimately close to God:

I saw the Lord, high and exalted, seated on a throne … Above him were seraphim, each with six wings: With two wings they covered their faces, with two they covered their feet, and with two they were flying.

Of course, angels might not exist, and the task of defending their reality is beyond our aim (e.g., Bazán 2010; Doolan 2012). The exact role of angels in our proposal will be further elucidated. At this moment, it is safe to say that within Christian theology, angels are spiritual agents with expert knowledge of God.⁵ Let us proceed within this Christian perspective.

Given the Christian theological framework, perceptions of God are verifiable in the angelic community. With superior knowledge of God, angels are supposed to be able to discern genuine experiences of God from those triggered by artefacts or devils. Thus, the conditions in SV are satisfied for perceiving God when angels are included as relevant agents in the picture. Possible evidence exists, for angels, to properly verify divine perceptions. We cannot but speculate what such evidence precisely is and shall not elaborate a theory of angelic evidence. Nevertheless, assuming that angels possess superior knowledge of God, we can generally regard them as having access to sufficient evidence about when God is perceived. Surely, angels might not eliminate all possibilities of error vis-à-vis perceptions of God, and they could lack non-circular proof of their own epistemic veracity. However, verification is not expected to serve the anti-sceptical aim of eliminating all possible errors or providing non-circular proofs. Under externalist principles, it suffices that angels can reliably perceive God and verify such perceptions. Another potential objection alleges that angelic perceptions of God are direct, rather than evidentially mediated, so they cannot satisfy the conditions in SV. In response, evidence is not necessarily incompatible with immediate knowledge: for example, on certain readings of Descartes, knowledge of one’s cogito is direct, but it still rests on the clarity and distinctness of its truth (e.g., Carriero 2009). More importantly, instead of evidence, SV may construe verifiability in other terms, such as cognitive processes to check, or even knowledge of, the veracity of relevant perceptions. What we need is just that angels have access to the veracity of divine perceptions, not that such access is evidential.

What about SCP? From the angelic view, are there sufficient conditions for perceiving God? As humans, we are ignorant of such conditions because, first, we do not know how reliably each level of spirituality facilitates perceiving God, and second, we do not know when God intends self-disclosure. Hence, there might exist conditions sufficient for perceiving God even from the human perspective; we just do not know what they are and when they are satisfied. In both aspects, angels are more knowledgeable. As messengers of God, angels should know how reliably each level of spirituality facilitates the perception of God. As superior perceivers of God, they can also know when God intends self-disclosure. Consequently, SCP’s conditions are properly satisfiable from the angelic perspective.

One might argue that God already possesses these very same features. The omniscient God knows when a mystical experience is genuine and thereby meets whatever conditions SV and SCP posit for divine perceptions. Arguably, God also satisfies the conditions in SV-PS as an ‘other agent’ who can verify a mystic’s divine perception. Why, then, do we need angels?

There is a limit, however, to how far God performs social verification. Strictly speaking, God does not exemplify the plural ‘other possible agents’ in SV-PS. More importantly, epistemic communities require members to share similar cognitive abilities. Since God is uniquely infinite, He is not a member in any community with finite beings, thereby precluding the A-social condition in SV-AS. This is why, to properly satisfy social verifiability conditions, we need angels more than God. Angels are not isolated from each other. They form a heavenly community wherein knowledge about God is shared. According to Aquinas, a ‘higher angel knows more in God about the meanings of God’s works than a lower angel does and … enlightens the lower angel’ (2006b, 95; ST I: q. 106: a. 1). Therefore, higher angels’ impeccable knowledge of God can disseminate, rendering divine perceptions verifiable in the angelic community. P-social verifiability is satisfied because perceptions of God – by humans or angels – can be sufficiently verified in the possible angelic community. A-social verifiability is also satisfiable because when one angel perceives God, it can be verified by other angels in the same group. Naturally, insofar as angels ‘always’ perceive God (Matthew 18: 10), they are unlikely to engage in the activity of verifications; the need for verification is arguably absent from the angelic realm. Still, as argued, verifiability is not about social practice. It concerns evidence, cognitive methods, or knowledge, available to other agents. Even if angels never verify, divine perceptions can still be verifiable for angels.

To summarise, within Christian theology that posits angelic communities, ‘direct perception of God’ is socially verifiable according to both SV-PS and SV-AS. Angels also have epistemic access to the conditions sufficient for perceiving God, as formulated in SCP. Surely, this proposal conflicts with theological positions that deny angels or angelic knowledge of God. Rather than entering theological debates, we may simply rest our view on those doctrines that favourably acknowledge angelic perceptions of God and their verifiability in angelic communities.

Multiple questions still await to be answered. Can we appeal to angels in epistemological discussions? How do angels affect human mystics? The remainder of this article develops our proposal by addressing three such issues. First, we explain why it is legitimate to incorporate angels within externalist accounts of divine perceptions. Second, we argue that angels are relevant to humans because both instantiate the same general type of ‘direct perception of God’. Third, we establish that human mystics and angels can be included within the same heavenly community.

Incorporating angels in the conditional externalist approach

It is viciously circular to use Christian doctrines to defend the justificatory power of mystical experiences and use these experiences to justify Christian doctrines. Is our appeal to angels similarly problematic?

Addressing this concern requires clarifying our externalist stance. Externalists often design hypothetical models about the essence of epistemic states. Providing evidence for such models is inessential to these projects. Reliabilism, for instance, construes knowledge as reliably produced true belief. This thesis does not assert that our beliefs are indeed true or reliable. Instead, it claims that a belief is knowledge if it is true and reliably produced. Our reliability, or the evidence thereof, does not bear on the truth of reliabilism. As Alston (1991, 103) remarks, reliability is not provable in a foundational fashion: even if the world is real and our cognitions reliable, any attempt to show this will be epistemically circular. At most, reliabilists can assert the conditional claim that ‘if sense perception is reliable’, we can use it to ‘show that it is’ reliable (Alston 1991, 148). Hence, devising a hypothetical epistemic model is one thing; establishing the reality of that model is another. This distinction permits distinguishing two reliabilist claims.

Reliabilism_Essence: An agent knows p if her belief p is true and reliably produced.

Reliabilism_Ascription: Moore knows that he has hands [because Moore’s belief that he has hands is true and reliably produced].

Reliabilism_Essence outlines the nature of knowledge. Reliabilism_Ascription ascribes reliability to actual agents. Although our reliability is evidenced to some degree, we cannot demonstrate it without presuming the reliability of the employed methods. Thus, Reliabilism_Ascription presumes the bracketed contents: beyond the conditional claim that Moore knows if he is reliable, it further assumes that he is reliable. For Alston (1991, 10), this is ‘a fundamental presupposition’ in cognitive practice. Let us say that Reliabilism_Essence makes a conditional assumption about the nature of knowledge, whereas Reliabilism_Ascription makes a factive assumption about the actual world.

The same distinction applies to mystical experience. Before making assertions about actual mystics, we should theorise what mystical experiences are and whether they offer evidence under the hypothetical conditions. Consider Alston’s account of mystical doxastic practice (MDP). On Alston’s view, mystical experiences offer justification if they originate from a socially established doxastic mechanism that is (a) self-sustaining, (b) not proven unreliable, and (c) subject to overriders. Dismissing verifiability as inessential, Alston insists that doxastic practices are not necessarily (d) socially verifiable. Conditions (a) to (c) define doxastic practice regardless of its occurrence in actual world. Parallel to Reliabilism_Essence and Reliabilism_Ascription are the following assertions:

MDP_Essence: A religious belief is justified if it is a product of a mystical doxastic practice, i.e., a product of divine perception via a socially established belief-forming system that is (a) self-sustaining, (b) not proven to be unreliable, (c) subject to overriders, but not necessarily (d) socially checkable.

MDP_Ascription: St. Teresa’s religious belief that Jesus is present is justified [because it is a product of a mystical doxastic practice].

MDP_Essence pertains to the nature of mystical doxastic practice; MDP_Ascription ascribes the practice to actual mystics. This distinction helps pinpoint the target of disanalogy objection. When objectors disregard mystical experience as unverifiable, they are attacking MDP_Essence rather than MDP_Ascription. The objection is less about actual features of mystical experience than its potential. To counter the objection, therefore, we need to restore the verifiability condition in MDP_Essence.

To fortify MDP_Essence, note that conditions a)–c) represent only ‘internal’ facets of doxastic practices, not the ‘external’ conditions where these practices operate. Typically, externalists make conditional assumptions about both internal faculties and external conditions. In Reliabilism_Essence, beliefs are conditionally assumed to be reliable, but reliability requires a physical environment. Hence, physical reality already underlies the conditionally assumed reliability of sense perceptions. Mirroring this construal of sense perception in a physical world-view, mystical experience can also be conditionally modelled in a religious world-view. Thus, conditional assumptions about mystical experience may already involve religious truths, such as God. Surprising as it appears, this aligns precisely with Alston’s approach:

I will be concerned only to argue that if God exists it is a real possibility that experiences like the ones under consideration constitute genuine perceptions of Him. (Alston 1991, 54, emphasis added)

He supernaturally brings about the requisite experience … If God exists, an assumption under which this discussion is being conducted, this is a possibility that cannot be ignored. (Alston 1991, 60, emphases added)

Rather than using mystical experiences to defend the existence of God, Alston conditionally assumes the religious truths. Indeed, to explore the epistemic potentials of mystical experiences, we should consider their optimal environments even if these environments are not real.

How do angels step into this picture? In positing God, MDP_Essence is already conditionally assuming the Christian world-view. Angels, along with God, are integral to this Christian perspective. We can thereby articulate angels who are already in this religious backdrop. Compared to MDP_Essence, introducing angels incurs no extra burden. Even if mystical experiences are unverifiable for humans, we do not need this restrictively human perspective when hypothesising the optimal conditions for mystical experiences. Because perceptions of God are socially verifiable for angels, we can revise the MDP assertions as follows:

MDP*_Essence: A religious belief is justified if it is a product of a mystical doxastic practice*, i.e., a product of divine perceptions via a belief-forming system that is a) self-sustaining, b) not proven to be unreliable, c) subject to overriders, and d) socially verifiable.

MDP*_Ascription: St. Teresa’s religious belief that Jesus is present is justified [because the belief is a product of a mystical doxastic practice*].

Mystical doxastic practice*, with angels in the assumed background, features a verifiability condition that effectively addresses the disanalogy objection. Because angels are conditionally assumed, MDP*_Essence does not require evidence for their reality. Moreover, similar to Alston’s exemption from debates about God’s existence or challenges to God’s essence such as the problem of evil, we need not defend the existence or metaphysical possibility of angels.

Is our improved model ascribable to actual mystics? Mystics empirically demonstrate the features in a)–c), but since we lack evidence for angels, are we still rationally entitled to ascribe mystical doxastic practice* to humans? If not, it appears that proposing MDP*_Essence would be an intramural theistic exercise detached from human activities. Does upgrading MDP*_Essence compromise the rationality of MDP*_Ascription?

The question of how actual humans are related to angels will be further elaborated in subsequent sections. Here, suffice it to note that the criticism involves a double standard. Our idea that Moore’s perceptual belief about his hands is reliable depends on the factively assumed reliability of sense perceptions through which we observe Moore’s cognitive conducts. We cannot avoid using sense perceptions when making assertions about actual sense perceptions themselves. By parity, mystics are rationally entitled to use perceived contents in mystical experiences as evidence about what their mystical experiences actually are (see also Losin 1987; Webb 2020). Angels, in particular, are frequently perceived in spiritual encounters: Isaiah saw a seraph (Isaiah 6:6–7), Joshua encountered a commander of God’s army (Joshua 5:13–14), and Ezekiel had visions of cherubim as winged creatures (Ezekiel 10:21). Unlike characteristically non-sensory experiences of God, angels often manifest with visible attributes. Despite this disparity, both angels and God can be classified as contents of mystical religious experience.⁶ Like various sensory processes, such as vision and touch, that collectively offer evidence about sense perceptions, so can perceptions of angels and God provide evidence about mystical experiences themselves. Given such evidence about angels, mystics can rationally regard their divine perceptions as a type of state that is exemplified by angels and thereby ‘verifiable’ in the angelic community. MDP*_Ascription gains similar evidential support as the ascription of reliability to Moore’s perceptual belief.

We can now clearly articulate how our solution diverges from Alston’s approach. Alston argues that if God exists, many mystical experiences are genuine and capable of providing epistemic justification. In response to the objection that mystical experiences lack prima facie justificatory value due to their non-verifiability, he posits that such verifiability is unnecessary, dismissing its requirement as ‘epistemic imperialism’. Although Alston’s position has faced criticism, our aim is not to challenge its cogency. Instead, we propose that there is no need to invoke epistemic imperialism. Contrary to the disanalogy objection, mystical experiences are verifiable by angels, and angels can be integrated into the debate in two ways. First, if we conditionally assume the reality of God as Alston does, it is a natural extension to also conditionally assume the existence of angels. Under this assumption, mystical experiences can be verified by angels. MDP*_Essence holds true. Second, mystics have experiences of angels. Alston maintains that it is legitimate to use contents received from sense perceptions as background information about those perceptions themselves. By parity, it is equally adequate for mystics to reference angels perceived in their mystical experiences when interpreting those experiences. Consequently, we can ascribe verifiable mystical experiences to actual mystics as illustrated in MDP*_Ascription.

Beyond Alston’s doxastic practice approach, our angelic solution applies to other externalist theories. The disanalogy objection might claim that mystical experiences, being socially unverifiable, lack epistemic values even if they satisfy externalist conditions such as safety (Broncano-Berrocal 2014; Grundmann 2018), sensitivity (Nozick 1981), or manifestation of epistemic virtue (Sosa 2021). Introducing angels is an effective response to such objections. In addition to these externalist conditions, mystical experiences are also verifiable within the full conditionally assumed religious background where angels are included.

The general ‘direct perception of God’ by angels and humans

To take stock, we have argued that perceptions of God are verifiable by angels within Christian theology. It is appropriate to reference angels through the externalist hypothetical-conditional approach to the essence of mystical experiences. Mystics can also rationally ascribe such angelically verifiable divine perceptions to themselves based on their available mystical-perceptual evidence about angels. More specifically, within the presumed Christian world-view, angels’ perceptions of God are not only P-socially but also A-socially verifiable in their own community. For humans, mystical experiences are P-socially verifiable – by angels as ‘other possible agents’ – and they can rationally ascribe such P-socially verifiable mystical experiences to themselves.

A significant objection alleges that humans and angels are fundamentally different and thus cannot share the same type of divine perception. The social verifiability of one does not transfer to that of another due to disparate underlying mechanisms. More broadly, objectors could challenge the idea of defending the justificatory power of human mystical experience in light of angelic perceptions. For instance, the excellent echolocation ability of bats does not mean that humans can use echolocation effectively. Humans may use echolocation, but our methods are essentially different and less reliable than those of bats. Given the profound differences between humans and angels – arguably greater than those between humans and bats – how can they instantiate the same divine perception with similar justificatory force? Here, one might accuse us of improperly leveraging the ‘generality problem’, that is to say, the idea that there is no principled standard for choosing the level of generality to describe cognitive processes, thereby allowing all process-tokens to be deemed reliable (e.g., Beebe 2004; Conee and Feldman 1998; Lyons 2019; Matheson 2015). Consider the vision of a myopic person. Such vision is obviously unreliable, especially for discerning objects that are small or at a distance. Nevertheless, if we classify this vision as ‘perception’, it will be unduly considered ‘reliable’ simply because perceptions are normally reliable. Mystics are like ‘myopic’ perceivers of God. Compared to angels, they are unreliable for perceiving the spiritual world. However, if we classify human mystical experiences as ‘direct perception of God’ and assume their reliability solely because angels are reliable in such perceptions, we risk mischaracterising human mystical experiences.

To these objections, our first response is that even if human and angelic perceptions essentially differ, human mystical experiences remain P-socially verifiable by angels. Because angels possess superior knowledge of God, they are in an epistemic position to verify human experiences of God even if these experiences are distinct from their own.

Additionally, objectors should not assume that mystical experiences are ‘unreliable’. Human mystical experiences are sporadic, but that does not entail unreliability. To evaluate the reliability of mystical experiences, we should ask how likely a mystic genuinely encounters God when she has a mystical experience about God, rather than when she tries to have such experience. Scarcity is not an indicator of unreliability. Objectors might insist that human mystical perceptions are unreliable, but this is merely an assumption and raises a different issue than the disanalogy objection. Our aim is to defend the social verifiability of mystical experiences against the disanalogy objection, not to demonstrate their reliability.

To establish angels’ relevance to human mystical experiences, we need more than their P-social verifiability by angels. We should further argue that humans and angels do share the same general type of ‘perception of God’. This proposal does not entangle us in the generality problem. The generality problem rests on the indeterminacy of selecting descriptive levels. It contends that we should not take myopic persons as using ‘perception’ where a more specific description is needed to indicate their unreliability. For experiences of God, we are exempt from this problem because they have a determinate description: be it for humans or for angels, it is just ‘direct perception of God’. This is the standard definition of angelic perceptions of God as in Pace’s view; it is also the prevalent account of human mystical experiences. Humans and angels both instantiate the same general ‘direct perception of God’.

Angelic perceptions of God are typically considered ‘direct’ because they do not need mediation by sense or intellect. For human mystics, such directness has also been defended. Alston, specifically, considers the following objection from Garrigou-Lagrange (1937, 270):

St. Thomas states explicitly and proves that no vision inferior to the beatific vision can make us know God as He is in Himself … All mystics tell us that they perceive, not God himself, as He is, but the effect of His action on their souls.

In response, Alston (1991, 61) emphasises that mystics often report ‘that God Himself appeared to their experience, not that they were only aware of His effects’. Furthermore, Alston argues that perceiving the effect of an object can be a way to perceive the object. When a person is seen ‘on television’, she is directly perceived since she is ‘identified with an item’ in the visual field, rather than through a different object (Alston 1991, 21). In this respect, when mystics experience God as gracious, their perceptions are direct because they do not involve objects other than God.⁷ Regarding Garrigou-Lagrange’s appeal to Aquinas, we acknowledge that humans lack access to full beatific vision, but this is compatible with directly perceiving God. According to Aquinas, humans are limited in their perceptions of God due to the sensory bases of their faculties: ‘Our souls, so long as we are in this life, have their being in corporeal matter, hence they cannot by nature know anything except what has its form in matter or what can be known through such things’ (2006a, 37; ST I: q. 12: a. 11). Still, it remains possible to directly perceive God via non-sensory mystical experiences. Aquinas hinted at this possibility when he recognised that ‘the more the soul is abstracted from material things the greater capacity it has for understanding abstract intelligible things. Thus divine revelations and foresights of the future come more often during dreams and ecstasies’ (2006a, 38; ST I: q. 12: a. 11). Surely, human and angelic perceptions of God are saliently distinct: the angelic beatific vision encompasses the ultimate bless beyond worldly humans; one may also assert that angels see God’s essence that extends beyond human comprehension. These differences, however, mainly concern ontological-spiritual properties of humans and angels as well as the contents of their perceptions. They are not necessarily epistemological.

Vis-à-vis the general idea that humans are metaphysically separate from angels as they are from bats, we suggest that the human-angel ontological divide is less significant than it appears. The human-bat chasm is unbridgeable: humans cannot metamorphose into bats or naturally adopt bat-like echolocation. In contrast, the human-angel divide is less rigid, especially in light of humans’ spiritual destination in Christian theology. In explaining why the pure in heart (mundum cor) perceives God, Augustine referred to angelic paradigms:

… when the only-begotten Son … declares Him with an indescribable utterance, the rational being, pure and holy, is filled with the indescribable vision of God, which we shall attain when we have become like the angels … (Augustine 2008, 191; Ep. 147. 22)

Therefore, despite human cognitive limitations regarding the full essence of God, two substantial links connect humans with angels. First, human perceptions of God could be explanatorily dependent on angelic models. Because humans typically perceive through bodily eyes, to which God is invisible, mystical experiences should be interpreted with non-corporeal models as epitomised by angels. This is why humans perceive God’s substance when ‘divinely rapt from this life to the angelic life’ (Augustine 2008, 199; Ep. 147. 31). Second, humans can strive towards this elevated comprehension by purifying their hearts. Christianity views human life as a journey to ‘become like the angels’ (see also Lootens 2012, 66). Human and angelic perceptions of God are accordingly interconnected in a static and dynamic sense. Statically, human mystical experiences are explainable through the lens of angels. Dynamically, humans have the potential to develop angelic divine perceptions. These links distinguish the human-angel relationship from the human-bat relationship: we neither explain human perception with bat perception, nor are humans disposed to become like bats.

Preliminary members of the heavenly community

Even if mystics and angels share the general ‘direct perception of God’, critics may still consider human mystical experiences as lacking in verifiability. While referencing angels makes human mystical experiences P-social verifiable, they still fall short of A-social verifiability. Objectors might contend that this lack of A-social verifiability is not accidental but rather an inherent limitation of human cognition. Mystical experiences are not A-social verifiable, not because humans happen to lack a community in which these experiences are properly verifiable, but because no earthly community can ever achieve the task.

We counter this objection by arguing that mystical experiences are indeed A-socially verifiable because humans can be regarded as members of a broader celestial community. For Christians, aligning humans with angels in the same community is a familiar concept. The Augustinian ‘City of God’ is a classic example where both terrestrial and celestial beings are integrated into the same collective narrative based on ‘love of God’ (Augustine 2013, 609). Furthermore, just as angels deliver divine messages, mystical prophets like Moses acted as intermediaries, conveying God’s words to humanity. These prophets played the same cognitive role of angels – as God’s messengers – within the general angelic-human community.

Critics might allege that these connections are insufficient for constituting a robust epistemic community. They might point out that the Augustinian City of God is grounded in divine love rather than epistemic functions, and that prophets are rare in Christian accounts of human history. What epistemic communities require are frequent communication and information sharing among the members. Because humans and angels rarely engage in such activities, they do not belong to the same epistemic community.

For the sake of argument, we concede that there is a properly restrictive notion of epistemic community and that mystics are not full members of such an epistemic community consisting primarily of angels. Nevertheless, we maintain that human mystics are novice perceivers of spiritual truths. They are accordingly preliminary members of the heavenly epistemic group.

To elucidate the notion of preliminary membership, consider small children in our community. Young children, still cognitively developing, may not yet grasp how to check their perceptual claims as adults do, but they are already justified given their reliable cognitive faculties. Proponents of SV-AS, to explain children’s justificatory status, must regard their perceptions as verifiable in a community to which they actually belong. Alston, when applying the doxastic practice approach to children, regards them as forming ‘a more primitive kind of perceptual practice than we actually have in mature human beings’ (1991, 160). It is true that we may take children as forming sub-groups within our general community. Meanwhile, children’s communities are not separate from our own. Epistemic communities are not necessarily homogeneous. They can involve complex structures with individuals of diverse cognitive proficiency: some members hold central roles, others are peripheral, and there can even be multiple centres with various topological features depending on their functions in the overall community. Children, in this respect, can be viewed as preliminary members of our community. This concept can be fleshed out in two perspectives:

The Static View: An agent is a preliminary member of a community if she occupies peripheral layers in the community.

The Dynamic View: An agent is a preliminary member of a community if she will normally become a full member of the community.

The static view defines membership according to an agent’s actual position in the group’s structure. While a detailed theory of ‘peripheral layer’ is beyond our scope, a sufficiently robust criterion is that static preliminary members (i) possess the same cognitive abilities as core members, (ii) perceive some of the same object-tokens as core members, and (iii) have been in contact with core members. Children satisfy these criteria: they share our cognitive capabilities, perceive the same physical realities, and are mostly in contact with adults. A man born and living alone on a remote island, instead, is not our static preliminary member because, despite possessing the same cognitive abilities, he does not perceive the same object-tokens as we do or interact with us. The dynamic view, in contrast, construes membership based on an agent’s prospective trajectory in the community. Beyond mere future possibilities, this view requires that the agent be expected to become a full member: her eventual central place must be warranted in the group. Most children satisfy this condition as they grow within our society and are supposed to become full members. These two views are non-equivalent. A static preliminary member may be unlikely to become a full member, thereby lacking dynamic preliminary membership, whereas a dynamic preliminary member may still lack the requisite cognitive ability and fail to be a static preliminary member.

Interestingly, mystics can be preliminary members of the heavenly community on both views. Although they are ignorant of how to verify divine perceptions, they perceive God similarly to children perceiving physical objects in our mature community. On the static view, mystics are preliminary members because they share the general ‘direct perception of God’ with angels, perceive the same truths about God, and have been in contact with angels directly through mystical perceptions or indirectly through angels’ words in the scripture. On the dynamic view, human mystics are preliminary members due to their eschatological hope for redemption: purifying their hearts, mystics are on a path towards a beatific state akin to that of angels. From both perspectives, mystics hold preliminary memberships in the broader heavenly community like children in our own. Thus, their mystical experiences are A-socially verifiable.

A potential issue is that children appear to differ from mystics in terms of parental guidance. Parents actively shape children’s perceptual world. They often correct misperceptions, such as when a shadow is mistaken for a ghost. In contrast, angels do not similarly check human perceptions of God. Consequently, the analogy between children and mystics seems amiss.

This objection assumes that guidance by core members is indispensable for dynamic preliminary membership. Yet, it is debatable how much guidance is required. Parents frequently care for children; angels also guide human spiritual life as messengers of God. There is a distinction, indeed, but why is the former sufficient for preliminary membership while the latter is not? Absent a principled theory of sufficient guidance, this objection poses no genuine threat.

Moreover, the objector overlooks that parental guidance is less about verifying children’s perceptions and more about shaping their conceptions. When mistaking a shadow for a ghost, a child is not suffering from perceptual inaccuracy but from a misunderstanding of physical surroundings. By clarifying that it is only a shadow, parents are refining her concept of the world rather than checking her perceptions. Mystics, mostly mature individuals, have outgrown the need for conceptual guidance. They are familiar with the physical world and have a cultivated notion of the spiritual domain through religious education. Although mystics might still struggle to discriminate genuine divine perceptions, they are not ignorant of the religious world in the same way as a child mistaking a shadow for a ghost. As things stand, angels do hold significance in informing humans about the religious world. They deepen human understanding of religious truths by elucidating visions (Daniel 8: 15–17), announcing the resurrection of Jesus (Matthew 28: 5–7), and prophesying divine judgement (Revelation 14: 6, 7). Therefore, the objection falters not only on the threshold problem of how much guidance is needed for dynamic preliminary membership but also on the nature of the guidance. More precisely, if the objection requires that preliminary members’ cognitive processes be checked by core members, then even parents do not universally meet this standard. If, instead, it requires that preliminary members receive general guidance from core members – be it cognitive or conceptual – then angels fulfil this role to a significant extent. Either way, both mystics and children are well positioned in their respective epistemic communities.

Conclusions

This article presents a novel externalist response to the disanalogy objection. By incorporating angels within the hypothetical-conditional religious settings crafted to assess the epistemic potential of mystical experiences, these experiences become verifiable in a society to which humans belong as preliminary members. Angels are often neglected in contemporary analytic philosophy of religion, arguably because they are neither part of the natural world on which analytic philosophers prominently focus nor a central topic in the legacy of the philosophy of religion. Nevertheless, the concept of angel has historically been instrumental in probing key philosophical topics (e.g., Hoffmann, 2012). Our proposal aligns with this tradition. We do not defend the reality of angels, nor do we prioritise externalism over internalism. Rather, our thesis is that the externalist framework, despite its usual naturalistic orientation, is compatible with incorporating angels to build a model of mystical experience that is socially verifiable. Beyond Christianity, this solution to the disanalogy objection extends to other religions, such as Buddhism and Daoism. Although these traditions might not recognise the Christian concept of angels, our solution is applicable insofar as they posit spiritual beings who are more knowledgeable about ultimate realities and form a higher spiritual community that humans can eventually join.