Questioning how religious experience is possible highlights an essential feature of Kantian critique. Being critical, in Kant’s sense, requires thinking perspectivally, identifying what is good or true about opposing sides of a controversy and distinguishing these from what is bad or false about each side. Kantian criticism typically focuses on how some feature of human experience is possible, detecting formal/subjective conditions that necessarily ground its possibility. Kant calls such necessary conditions for possibility ‘transcendental’. However, not all types of human experience can be traced back to a fully transcendental ground. Taking Kant’s view of religious experience as a case in point, I shall raise three questions that arise out of the recent literature on Kant and religion.

First, what role (if any) does Religion within the Bounds of Bare Reason Footnote ¹ play in Kant’s philosophical system? Many interpreters have assumed the book is primarily an appendix to his ethics – a mere supplement to the system’s practical standpoint, not a constituent component of doing critique. For example, this traditional assumption seems to operate tacitly throughout much of Allen Wood’s Kant and Religion (Wood Reference Wood2020; citation throughout by page number only), for Wood frequently interrupts his exegesis of Religion to insert sometimes lengthy explanations of Kant’s ethics. However, Wood also presents an up-to-date account of the crucial role symbolism plays in Kant’s theory of religion. Early in chapter 1 he advances a thesis that will surprise some readers, that Kant’s Religion is largely about religious experience. The first Critique’s limitations on cognition of God, he insists (p. 5), ‘do not preclude what might be called – and even what Kant himself might call – an “experience” of God’. Although such experience conveys no ‘theoretical cognition’, we can ‘accept [it] on practical grounds or as part of aesthetic experience. Such experiences are symbolic in character.’Footnote ² This opening argument suggests that for Wood Religion is a sequel to the third Critique, not the second.Footnote ³ This way of positioning Religion within Kant’s corpus fits best with the claim Wood develops quite effectively in his book’s second half, that symbolism is of crucial importance to Kantian religion.Footnote ⁴ Yet the book’s first half reads Kant’s Religion as if it were simply his system’s third book on ethics. This tension illustrates why it is so important for interpreters to take a clear stand on whether Religion is a proper component of the critical system at all, and if so, to explain what role it plays therein.

My second question is this paper’s title: how is religious experience possible? I shall consider this and my third question using Wood’s aforementioned book as a sounding-board, as it aptly illustrates some key aspects of the relevant controversies. The previous quotation states Wood’s main answer to this second question: for Kant, symbolism makes religious experience possible. Wood repeatedly associates symbolism with the very possibility of religious experience (p. 151): ‘To say that the life, death, and resurrection of Jesus Christ symbolizes something about the moral change in ourselves is not to say that we reject what is presented through the symbol. … For the symbol is what gives us experiential access to what it symbolizes.’ Religious experience would therefore be impossible without the symbols provided by some historical faith.Footnote ⁵ If symbolism serves for Kant as a necessary condition for the possibility of religious experience, this gives rise to the question: does the possibility of religious experience depend on any necessary conditions other than the presence of properly functioning symbols? Wood rightly emphasizes that Kant’s ‘save’ for religious experience, which interpreters typically read Kant as altogether forbidding, is to regard the moral law as ‘inner revelation’ (p. 18) – a move fully consistent with Kant’s restrictions on theoretical cognition, as long as we interpret it symbolically rather than literally. But are there other conditions?

A prime candidate for such a further condition is Kant’s theory of the ‘archetype’ of perfect humanity, sometimes virtually identified with the ‘good principle’. In Religion’s second piece, Kant regards this idea, that human nature is grounded in a being who perfectly fulfils all levels of our predisposition to good (i.e. animality, humanity and personality), as the only hope we have of experiencing moral improvement, given the debilitating effects of our propensity to evil. Kant never presumes to argue for the archetype’s reality; he merely takes as granted that, if our nature is beset by an evil propensity, the only rational way out would be to appeal to God’s gracious provision of such an archetype conveyed by reason. (Kant’s portrayal of the archetype as explaining the possibility of any alleged experience of grace closely parallels the role of possibility in his moral arguments for God’s existence.) For Kant, morality itself cannot have a transcendental status, as he reserves this term for arguments proving a necessary condition for the possibility of theoretical cognition. Nevertheless, Kant’s frequent appeal to the possibility of realizing various practical objects, and his claims that such possibility can be explained only by presupposing that some subjective conditions necessarily apply, have led me to call various practical arguments ‘quasi-transcendental’. A quasi-transcendental argument follows a structure similar to Kant’s standard transcendental arguments but takes as its object something other than an empirical cognition, such as a religious symbol. My second question can therefore be restated as follows. Aside from the basic need for symbols conveyed by some historical faith and for an archetype of perfection that can give us hope of achieving moral improvement, what other quasi-transcendental conditions for the possibility of religious experience does Kant’s Religion introduce?

In answering this refined second question, some interpreters might cite Kant’s theory of humanity’s ‘unsociable sociability’ as another necessary condition for religious experience. Kant undoubtedly does present his theory of the ethical community as a necessary condition for authentic religion. Moreover, on the basis of his new, ‘religious argument’ for God’s existence,Footnote ⁶ he also argues that, for us unsociably sociable humans, such a community can have its proper effect only by taking the form of a church. Thus, Kant argues that the existence of some historical faith or other is a necessary condition for the possibility of religious experience.Footnote ⁷ However, Kant never treats unsociable sociability itself as such a condition, though our proper response to it is an underlying theme running throughout Religion. Wood sometimes seems to portray unsociable sociability as functioning quasi-transcendentally, but in all such cases, the source of this status is the evil propensity.Footnote ⁸ This observation leads directly to my third question.

Could Kant’s theory of unsociable sociability explain the origin of human evil? Kant obviously regards unsociable sociability as the empirical product of our propensity, its manifestation in human history, not its source. Yet Wood reiterates the claim he had previously defended in many publications, that Kant grounds human evil in our need to exist in society (see n. 11). Obviously, our social nature is extremely important to Kant – important enough that in Religion’s third piece he makes the church a necessary condition for successfully combating the evil that grafts itself onto this aspect of our nature. But does acknowledgement of the inevitably social manifestation of our evil nature exclude the claim that Kant argues quasi-transcendentally when discussing the human propensity?

Wood summarily rejects all claims that Kant advances an ‘a priori’ argument for humanity’s evil propensity, oddly claiming (pp. 84–5) that ‘Kant’s thesis is … far too strong for it to be proven a priori’.Footnote ⁹ Unfortunately, he gives no details about these alternative interpretations or how they differ from each other.Footnote ¹⁰ Instead, ignoring such counter-evidence, he claims that the first piece passage commenting most directly on whether we can prove human nature’s evil status is one near the end of the Introduction, where Kant says ‘this can be proved only later on’, through ‘anthropological probing’ (6: 25). This passage appears before Section I, where Kant will defend the first premise of his quasi-transcendental argument that human nature is radically evil. Wood leaves unmentioned that in the context of this preliminary remark Kant explicitly states that the proof referred to here is not the thesis of the first piece, that human nature is evil, but the thesis of the second piece, where Kant first introduces the archetype of perfect humanity and argues that we have a duty to make this ideal of goodness objectively real. Wood’s favourite proof-text does not even address the question of whether we can prove human nature is evil; it addresses the question whether there is any ‘basis for exempting any human being from’ whatever the first piece ends up concluding about human nature.Footnote ¹¹ Anthropological concerns are important for Kant, but not in the first piece, which aims to determine how experiencing evil is possible; only in the second piece does Kant seek to determine whether ‘one individual could be assumed to be good, and another to be evil, by nature’, even if the entire ‘genus’ is evil (6: 25). Elsewhere, I refute Wood’s way of interpreting this passage in detail.Footnote ¹²

Wood concedes what my interpretation claims for Kant’s quasi-transcendental argument when he writes (p. 88): ‘This consequence of the thesis of radical evil for moral practice [that we need a change of heart in order to improve] is what is essential to the overall argument of the Religion.’ This comment offers hope that, if Wood grasped what my position on Kant’s quasi-transcendental mode of argument is, he could accept it. My claim is just as conditional as Wood’s: if religious experience is possible, one of its necessary conditions is that the human race has collectively ‘chosen’ evil. Wood here (p. 88) grants that this ‘consequence of the thesis of radical evil … is essential to the overall argument of the Religion’. Yet if so, it makes no sense to deny any a priori basis for the claim that ‘we cannot begin our strivings for moral improvement with a presumption of our innocence, but must presuppose an original depravity of our disposition, and therefore we need to combat it’ (p. 89). What do Wood’s ‘cannot’, ‘must’ and ‘need’ mean, if not that reason constrains us to think of evil thus – i.e. that there is something quasi-transcendental about the reasoning? Wood has no problem conceiving and accepting Kant’s view of human goodness – i.e., the moral law – as a priori (see e.g. p. 90). The point of Kant’s doctrine of radical evil is that we initially make ourselves rational by giving ourselves an upside-down version of the same moral principle. We must have done so, despite having no empirical evidence of that original act’s occurrence, because we observe the reversed principle operating from our earliest decision-making and because evil deeds could not exist, had we not chosen such a fundamental maxim. Wood cannot have his anthropological cake and eat it too: either the moral law itself is, like the propensity to evil, merely an anthropological observation that awaits empirical confirmation, or else both the moral law and our evil propensity can be established a priori for the species, although we cannot identify their relative presence in any given individual without ‘anthropological probing’ (6: 25).

In short, Wood (Reference Wood2020: 86) correctly reads this passage as making ‘unmistakably clear … that Kant thinks the thesis … has not yet been proven at all, much less proven a priori’. But Wood portrays Kant’s explicitly preliminary side-comment as if it were intended as the final answer to a question Kant has not yet started discussing in the first piece! In this passage, Kant portrays the claim that some special human beings have escaped the evil propensity as an anthropological question that, as such, does not properly belong to the first piece. Wood is therefore entirely unjustified to take this early comment as a reason for questioning (p. 86) whether ‘Kant himself is even committed to the thesis of radical evil’. (What Kant is not committed to, and carefully avoids appraising even in the later parts of Religion, is whether or not the historical Jesus possessed radical evil.Footnote ¹³ ) If Kant were as uncommitted to radical evil as Wood claims, then Kant’s explicit statements throughout the first piece, that human nature is radically evil, would have to be prevarications – a position Wood elsewhere resoundingly rejects.Footnote ¹⁴

On the basis of Kant’s claim at 6: 25 that the thesis of humanity’s radical evil ‘has not yet been proven’, Wood (p. 61) infers that Kant never gives any proof and that he therefore might not be ‘fully committed to the thesis that there is a radical propensity to evil in human nature’. But Wood can maintain this speculation only by carefully avoiding any close examination of Kant’s text in the first piece, which repeatedly affirms, even in the title of Section III, that ‘The Human Being is Evil by Nature’ (6: 32). Admittedly, Kant also says in Section III that, given ‘the multitude of screaming examples that experience places before our eyes in the deeds of human beings’,Footnote ¹⁵ ‘we can spare ourselves the formal proof’ of the stated thesis (6: 32–3; emphasis added); but he never says he actually omits this proof. Contra Wood, Kant subsequently warns that these ‘experiential proofs … still do not teach us the actual make-up of that [evil] propensity and the basis of this opposition [to the moral law]’ (6: 35),Footnote ¹⁶ and a few pages later he clarifies (6: 39n.): ‘The actual proof of this judgment [i.e. of Section III’s title] … is contained not in this section but in the previous one.’ Although Wood cites Palmquist Reference Palmquist2008, he simply ignores the massive amount of textual evidence provided there, demonstrating the utter irrelevance of his position to a proper interpretation of the possibility of human evil. He admits at one point, however, that Kant’s inference to the evil propensity appeals to a ‘relative kind of apriority’ (p. 75); with a more charitable treatment of his interlocutors, Wood could present his position as referring to the same feature of Kant’s argument that I call quasi-transcendental.

Wood supports his social interpretation of evil’s source with a second quote (pp. 77–8), this time from the introduction to the third piece. Yet the lengthy passage from Rel, 6: 93–4, is not about evil’s origin; Wood gives this impression by inserting ‘[the radical propensity to evil]’, which never appears in Kant’s original. Kant starts the quoted passage by saying a person ‘can easily convince himself’ (6: 93) of the argument that follows – ‘that [evil inclinations] come to him not so much from his own coarse nature … as from human beings with whom he stands in relation or association’. The words ‘coarse nature’ refer to Kant’s argument in the first piece, that we are each responsible for our own evil choices. His point here is that we cannot merely blame other people’s annoying behaviour for our own poor choices. Wood shows his own self-confessed bias (acknowledged in his Preface) when he writes (p. 78): ‘to say that for Kant the radical human propensity to evil has a social and historical origin is only to report what Kant explicitly says’. Kant explicitly says that human beings looking for excuses and not wanting to place blame where it belongs, on the choices that have determined their own coarse nature, easily pass off responsibility for their evil propensity onto other people. The position imputed to Kant by supporters of the claim that Kant grounds evil in unsociable sociability is what Kant explicitly identifies as a deception that easily tempts our evil inclinations!Footnote ¹⁷

My previously elaborated alternative interpretation is that Religion’s first piece provides a quasi-transcendental argument, whose main steps are easily overlooked until one notices that they are conveyed in the four section titles, which follow the form of the first Critique’s typical transcendental arguments. Kant’s proof is that, if evil exists, something in human nature must ground it, because beings such as ourselves, who possess a good predisposition, could not do evil if this were not the case; we clearly do make evil choices; therefore, the human agent’s volitional structure must have an evil propensity. Acknowledging the relevant textual evidence does not require abandoning the view that unsociable sociability is crucial for Kant; it simply requires us to concede that this doctrine tells us nothing about evil’s possibility; rather, it gives us a helpful way of describing evil’s impact on our attempt to implement the social (‘humanity’) aspect of our good predisposition. Given that Wood reads Kant’s Religion as generally focusing on conditions for the possibility of overcoming the effects of our evil nature through religious experience, he gains no advantage by continuing to insist that Kant took refuge in a merely empirical observation at this foundational stage of his argument, instead of conceding that the first stage of Kant’s argument employs his standard (quasi-)transcendental strategy.Footnote ¹⁸

Accepting the quasi-transcendental status of Kant’s argument for human nature’s evil propensity would not require interpreters to sacrifice anything crucial about the importance of unsociable sociability.Footnote ¹⁹ Reading Kant’s Religion as a systematically interconnected series of quasi-transcendental arguments for the possibility of religious experience keeps Kant’s doctrine of unsociable sociability firmly in place, not as the source of evil but as fulfilling the more plausible function of describing the paradoxes we face when considering the human situation empirically. Kant’s use of possibility-type proofs in so many other areas of his system – not the least being in his various moral arguments for God’s existence – is thereby rendered consistent with his position on evil, so that interpreters no longer need to portray Kant as being merely unable to make up his mind because ‘freedom is incomprehensible’ (p. 65n.).Footnote ²⁰ Wood’s alternative, that ‘The maxim of evil is grounded in a certain way of valuing ourselves – the comparative-competitive way’ (p. 78), lacks support from Kant’s texts and directly conflicts with numerous passages. The issue is not whether human nature inevitably exhibits this comparative-competitive feature; Kant states that explicitly in Section I of the first piece. Wood’s interpretative option deceptively downplays the fact that Section I introduces the predisposition to good. Kant’s main point there is that nature made us competitive for our own good, not that evil has its source in competitiveness. Interpreters will search in vain for the latter claim in Kant’s text, especially in Sections II–IV, which examine how we graft evil onto these good predispositions. Kant’s point in Section I is that our sociable nature is intrinsically good, precisely because it cannot (by itself) be the source or ground of the evil that makes us unsociable, much less of its very possibility. Moreover, recognizing the quasi-transcendental status of Kant’s general mode of argument in Religion offers a sound basis for understanding the profound role of symbolism in religion – a point that Wood, as we have seen, highlights excellently, but without bringing out its grounding in Kant’s overall argumentative strategy.

Our consideration of the consistency of upholding both a belief in unsociable sociability and the claim that Kant offers a quasi-transcendental argument for an evil propensity that first makes human evil possible, aptly illustrates Kant’s normal argumentative strategy. As a perspectival thinker, he claims both that evil has a quasi-transcendental grounding (in our necessarily presupposed propensity) and that evil manifests itself empirically in our unsociable sociability. Both views are correct, depending on which perspective we adopt. The debate over the nature of Kant’s thesis regarding the evil propensity seems intractable only when the interlocutors do not think perspectivally: the importance of anthropology for Kant should not be underestimated and is essential to his view of how symbols arise in any given religion; but it cannot be used to answer questions about what is most fundamental in philosophy – i.e. the possibility of things – because these are a priori questions.