Is Kant correct? Can we have no knowledge of God?

Reading Time: 13 minutes

By Bartlomiej Staniszewski, Oxford University

Disclaimer (13th of December 2021):

While this article offers some ways of escaping Kant’s critique, it does not evaluate those ways very thoroughly; I think it is most useful in inviting the reader to explore those which seem the most promising to the reader further.

I also have not given any space to what would be a Berkeley-inspired response to Kant – to deny the existence of noumena, on the basis that it makes no sense to talk about that which cannot be perceived by us. This is because it is an unpopular response, but I happen to think it is very plausible. A couple of years ago I wrote about some of my reasons for this (look for Immaterialism on my Academia page, For a modern (and very good) version of idealism, see John Foster’s A World for Us: The Case for Phenomenalistic Idealism.

Is Kant correct to claim that we can have no knowledge of God?

Kant is incorrect to claim that we can have no knowledge of God. This is because he fails at showing that transcendental idealism precludes cognition of God. In this essay, I will describe Kant’s critique of metaphysics as incapable of providing us with knowledge of God and its consequences for theology. I will then evaluate some criticisms of Kant’s critique to conclude that it is unsuccessful.

In the Critique of Pure Reason, Kant highlights that, as we cognise objects in space and time, we only cognise appearances (also called ‘phenomena’). However, we know nothing about the things in themselves (things as they are independent of our perception of them, also called ‘noumena’) of which the phenomena are appearances; cognition of phenomena can give us no knowledge of any noumena, which can be completely unlike the phenomena we cognise (Stang, 2021). But because we can only cognise phenomena, it seems that knowledge of anything outside of the scope of our experience (which is always in space and time) is impossible. This includes knowledge of God (Fugate and Pasternack, 2021).

Kant calls this set of doctrines ‘transcendental idealism’. He writes in the Critique that ‘if no intuition could be given corresponding to [a] concept, the concept would still be a thought (…) but would be without any object, and no knowledge of anything would be possible by means of it’ (Kant, 1966, pp. 85-86). He seems to think that we can only gain what he calls an ‘intuition’ regarding a concept via sense experience, as otherwise, the concept has no ‘object’, despite still being a thought – a thought that does not correspond to any external reality if it is not the result of sense experience. And indeed, it is difficult to explicate how one could cognise external reality absent of sense experience. It seems that one would have no guarantee that what one is thinking corresponds to external reality. A thought could be had regarding some entity X regardless of whether X has external existence or not. Therefore, mere thought divorced from sense experience seems to have no necessary connection to external reality (Grier, 2018). But cognition that is a result of sense experience has a necessary connection with the external reality of how things appear to us – with phenomena. Yet, since phenomena can be completely unlike phenomena, this means knowledge of noumena is impossible. As such, transcendental idealism seems true.

‘We cannot access the noumenal reality of God, so we cannot predicate anything of God’ besides how we experience God. We cannot even deny things of God, for then we would have to know that God is not-something, but we know nothing about noumena (Wolterstorff, 1998, p. 13). If transcendental idealism is true, we can never claim to know anything about, or even cognise, God. Kant’s solution, which has been more or less accepted by many modern theologians, is to only ever make practical postulates regarding God; what would be a useful, humane, or liberating conception of God (Plantinga, 2000, p. 5). Alternatively, we can make claims about the human experience of God. But natural theology (which makes inferences about God’s noumenal reality from knowledge of phenomena), and more generally, knowledge of God, is impossible, as are all claims which pertain to how things are outside of our experience. While the natural theologian may appeal to human experience of, say, causation, when articulating the Cosmological Argument, they can only understand how causation operates ‘in our experience, and there [is] modified and brought into relation to time [and space]’. Abstracted from our experience, we cannot specify what reality causation expresses (Walsh, 1939, p. 444). But when making truth-claims about God, a natural theologian wants to describe something that is true regardless of human experience; they want to make a claim which is necessarily related to its object, God. But transcendental idealism precludes those (ibid., p. 447).

Plantinga objects, however, that this is a more general problem with transcendental idealism, and not a special problem for theologians. For it is also true of all other things (not just God) that we can make no truth-claims about their noumenal reality. (Plantinga, 2000, p. 16) For example, transcendental idealism seems to also be a problem for science, as it seems to deny that scientific claims are necessarily related to their objects. But that is not a problem for the scientist. Kant does not deny that the objects studied by science are real; only that they are noumenal. The objects of scientific research, for Kant, are real phenomena, and scientific claims are necessarily related to said phenomena (Stang, 2021). Swinburne writes that were Kant aware of the discovery of unobservable entities posited by science, he may ‘have acknowledged great scope for human reason to acquire probably true beliefs about matters far beyond the observable’ (Swinburne, 2012, p. 329), but the unobservable entities posited by science are still observable by humans in principle, using appropriate machinery, like powerful microscopes or special lenses. The sentiment behind noumena is that they are unobservable from the human perspective even in principle. Hence, Kant ‘accepts the existence of unobservable entities posited by our best scientific theories and holds that these entities are [phenomena]’ and that phenomena can really ‘have properties quite different than they seem to have in sense perception’ (Stang, 2021). The properties that phenomena really have are those investigated by science, and how phenomena appear may be misleading to the way phenomena really are, but, despite this, the properties of phenomena investigated by science are observable by humans in principle, unlike the properties of noumena (ibid.). As such, transcendental idealism does not deny the reality of scientific claims, and it can accommodate how non-theological claims are necessarily related to their objects; their objects are phenomena. And if the object of my claim is not noumenal, I need not make truth-claims about noumenal reality. But as theologians, the object of our claims, God, is noumenal.

An influential critique of transcendental idealism is made by Strawson in his Bounds of Sense (1995, pp. 157-174). In positing the existence of noumena, Kant makes the following claims:

(E) There are noumena.

(N) Noumena are not in space and time.

But he also claims knowledge of noumena is impossible:

(H) We know nothing about noumena.

H seems to remove any warrant for E and N, for if we know nothing about noumena, we can hardly know they exist or are not in space and time. And if transcendental idealism renders itself unknowable, it seems to be a bad theory (ibid.). Admittedly, we might reject transcendental idealism while maintaining (H), but we would be hard-pressed to justify (H). Firstly, it is a better hypothesis where we posit fewer entities rather than more, and therefore, if we reject (E) and the justification for it, it seems that we do know something about noumena; they do not exist (Swinburne, 2012, p. 333). Furthermore, if we know nothing about noumena, it is not clear what it means to say we know nothing about them. By analogy, consider the claim (H*) We know nothing about nafoozle. How is (H*) to be understood if it is true? It seems to be impossible to understand, much like a paradoxical claim such as, ‘This statement is false’. But then it cannot be true. More so, assume that (H*) can be true. Nonetheless, you would not be justified in believing that (H*), precisely because you do not know what it means. But if so, you cannot be justified in believing that (H).

As a result of Strawson’s criticism, some Kant apologists argued that the phenomena/noumena distinction is not between two kinds of objects, but an adverbial distinction between two ways of considering the same object; as it appears, or as it is “in itself” (Stang, 2021). Allison writes that the distinction between phenomena and noumena is a distinction only in the perspective from which we consider things. We can consider them as objects of thought for human cognisers like us, in which case we are considering objects as phenomena. ‘Or we can abstract from our particular cognitive conditions and consider objects merely as objects for a mind in general, in which case we are considering them as things in themselves.’ (ibid.)

Van Cleve objects to Allison that the properties of a thing do not vary according to how it is considered (Van Cleve, 1999, p. 8). Compare how we might consider a job applicant; we may consider as abstracted from their race or sex. But this does not mean that we consider them to be race-less or sex-less. Rather, we are silent on the topic of those properties; we do not judge them to have any particular sex or race. But if so, then, by analogy, considering objects as abstracted from our particular cognitive conditions cannot mean we consider them as existing things outside of space and time. Rather, we refrain from judging whether they are existing and whether they are in space and time (Stang, 2021). But then, on Allison’s view, (N) and (E) are without support, and Allison encounters the same difficulties that transcendental idealism encountered as a result of Strawson’s critique.

Furthermore, Langton writes, Allison’s interpretation is compatible with the thesis that there are no noumena and any cogniser would cognise things just as humans do. In that case, even when abstracted from our experience, our understanding of categories like causality would hold, allowing the natural theologian to make their arguments. This would render transcendental idealism compatible with natural theology. But such an understanding loses what Kant is getting at – that ‘there is something about the world of which we are irremediably ignorant’ (ibid.).

As such, Langton interprets the phenomena/noumena distinction as between two different classes of properties had by objects: their extrinsic and intrinsic properties. Phenomena are objects qua bearers of extrinsic properties, and noumena are objects qua bearers of intrinsic properties. Therefore, she understands (E) as (E2): ‘Objects with intrinsic properties exist’ and (H) as (H2): ‘We cannot cognise the intrinsic properties of objects’. As ‘the having of intrinsic property’ is not itself an intrinsic property, E2 and H2 are compatible. (N) is also unproblematic for Langton; on her reading it amounts to (N2): ‘Being spatial is not an intrinsic property of objects’, and, since being spatial is an extrinsic property, N2 is trivial (ibid.).

But, Plantinga writes, for theism, such a reading is unproblematic. Aquinas and almost any other theologian would agree that ‘God, obviously enough, has many properties we don’t know about, and presumably many of which we could not so much as form a conception’ (Plantinga, 2000, p. 13). To say we cannot cognise His intrinsic properties is acceptable for a natural theologian. On such an analysis of religious language, any claim about God is a claim about His extrinsic properties and is still necessarily related to its object. As such, truth-claims about God can still be made. And while the inability to talk about God’s intrinsic properties may seem like a significant cost, it seems in accord with how many natural theologians write. For example, Anselm’s description of God as ‘that than which no greater can be thought of’ describes an extrinsic property of God, and attributes like omnipotence and omniscience can be understood, analogously, as ‘that than which no more powerful can be thought of’ and ‘that than which no more knowledgeable can be thought of’. Indeed, Aquinas holds that God’s essence is unknowable, and so, it seems he agrees that we cannot know God’s intrinsic properties, while Panchuk goes as far as to claim that God has no intrinsic properties (Panchuk, 2019, p. 231).

However, that is only so on what Stang calls the Identity Reading of Langton. On that reading, one and the same object (such as God) can have both intrinsic and extrinsic properties. Langton, however, repudiates said reading, precisely because it seems to give properties we can cognise to objects themselves. She instead endorses the Non-Identity Reading, on which phenomena are collections of the extrinsic properties of some objects. Thereby, extrinsic properties do not belong to any particular object (Stang, 2021). But this seems like a weird understanding. How can properties from different objects bond together to form one phenomenon? Furthermore, it seems to subscribe to a bundle theory of phenomena (analogous to a bundle theory of objects). And although Langton could assert that there exist some intrinsic properties of objects that explain both how extrinsic properties bond together across objects and how they have compresence in a phenomenon, that would be an ad hoc explanation. The Identity Reading seems more plausible.

Wolterstorff writes that we can avoid the problems caused by transcendental idealism in general by using definite descriptions to talk about God, for example, “Creator of the universe” or “The one who brought about all that might not have been.” (Wolterstorff, 1998, p. 18) But while transcendental idealism would allow for those descriptions to pick out some being, it would not allow us to have any knowledge of said being except for what we ascribe to it in the definite descriptions and may be unreflective of the noumenal reality of God. Furthermore, unless we produce sound arguments for the existence of God, we would not be justified to believe that said definite descriptions pick out any existing being, but transcendental idealism makes it impossible to produce sound arguments for the existence of God. Those problems only disappear on the Identity Reading of Langton.

Runzo, on the other hand, accuses Kant of evidentialism – the belief that for a belief to be epistemically justified, it must be supported by evidence (Runzo, 1991, p. 32). Runzo claims that theistic beliefs could be properly basic – ‘epistemically justified without evidence, or other beliefs as reasons’ (ibid., p. 33). Consider believing that you could not sleep last night. You do not need evidence or any other belief to be epistemically justified to believe that. You only need to have formed the belief under the right circumstances, that is, the circumstances of having not slept well last night. Analogously, Runzo writes, ‘a theist can properly believe, without inferring this from other beliefs as reasons, that “God created nature” or “God should be thanked.” For a theist often comes to hold these beliefs not on the basis of evidence or inference but immediately – e.g., while observing the works of nature, or while hearing sacred music’. (ibid., p. 34) We can gain knowledge of the noumenal reality of God immediately.

But it is unclear that observing the works of nature or hearing sacred music are the sort of circumstances that are the right circumstances to form the relevant properly basic beliefs under. Moreover, Runzo’s position is of limited help to natural theology, and is of little apologetic force. An atheist could analogously claim they have the properly basic belief that God does not exist, formed under the circumstances of observing evil, or they could simply lack the properly basic belief that the theist has, having not formed it themselves.

Similarly, one could not counter Kant by asserting that we have cognition of noumena from non-phenomenal sources. Kant explains in the Critique, discussed above, why neither mere thought nor phenomena can give us cognition of noumena. The only remaining options are that noumena themselves could cause cognition of them in us, or our cognition of some noumena could be pre-existing in our minds. But the point transcendental idealism drives is that we cannot know this – the thoughts that noumena cause in us, or the thoughts that are pre-existing in our minds, could be unrepresentative of the reality of the noumena. One way to avoid this problem is to assert that it is not noumena in general that cause cognition of them in us, but rather, God causes them, or at least guarantees that our thought coheres with the reality of the noumena. But this response begs the question. We cannot be justified in thinking God causes our thought to cohere with noumenal reality without first being justified in thinking God exists; justification transcendental idealism renders impossible.

A better strategy is to embrace epistemological externalism. For on externalism, we can know X without knowing that we know X. S’s knowledge of X is justified not internally, by S’s knowledge that S knows X, but by some facts external to S’s mind. A cogniser need not have any cognitive grasp of the facts which make their thought knowledge. For example, we would ordinarily hold that a falcon knows that they see a mouse in the field despite being unaware of what renders their thought that there is a mouse in the field knowledge (that is, unaware that they have extraordinarily good eyesight). But if externalism is true, we could know some fact Y about noumena even if we do not know how we know it. Still, the success of this strategy depends on the specific externalist account of knowledge embraced. For example, on Armstrong’s account, only non-inferential knowledge can be justified by facts external to the cogniser (Armstrong, 2000, p. 78), and it is difficult to understand how one could have non-inferential knowledge of noumena (perhaps aside from religious experience, which possibly may be used to justify further claims about noumenal reality).

Swinburne also argues that transcendental idealism is of limited power against abductive theology. For if we make a coherent hypothesis regarding God that predicts phenomena and is rendered probable by observable evidence, we have reason to suppose that the hypothesis is true (Swinburne, 2012, p. 328). But then, we have knowledge of a noumenon (God). But it is impossible for observable evidence to serve as evidence for the existence of noumena. For observable evidence is phenomenal, and we can never know what sort of noumenal reality phenomena represent.

More successfully, Swinburne highlights that it is false that we cannot cognise noumena. For we can abstract from our perspective sufficiently to have some understanding of what things may be like in themselves. For example, we can imagine taking a pill that abstracts our perspective from the category of space (as we understand it), placing us in a non-Euclidian plane, or giving us a six-dimensional perspective. But if we abstract from our perspective in this manner and arrive at the same conclusion regarding some object O in each perspective we consider, we thereby have some inductive guide as to what holds true of O regardless of perspective (Swinburne, 2012, p. 328).

To conclude, Kant is incorrect to claim that we can have no knowledge of God. This is because, on the most plausible understanding of the distinction between phenomena and noumena, we can still have knowledge of God’s extrinsic properties. Furthermore, on some externalist accounts of knowledge, we may have the potential to have knowledge of God regardless of Kant’s claims. And finally, we can have inductive knowledge of God if we can sufficiently abstract from our spatio-temporal perspective.


Armstrong, D. M. (2000) ‘The Thermometer Model of Knowledge’, in Bernecker, S. and Dretske, F. (eds.) Knowledge: Readings in Contemporary Epistemology. Oxford: Oxford University Press.

Fugate, C. and Pasternack, L. (2021) ‘Kant’s Philosophy of Religion’, in Zalta, E. N. (ed.) The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab: Stanford University Press.

Gardner, S. (1999) Routledge Philosophy Guidebook to Kant and the Critique of Pure Reason. London: Routledge.

Grier, M. (2018) ‘Kant’s Critique of Metaphysics’, in Zalta, E. N. (ed.) The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab: Stanford University Press.

Kant, I. (1966) Critique of Pure Reason. Translated by F. M. Müller. Garden City: Anchor Books.

Panchuk, M. (2019) ‘The simplicity of divine ideas: theistic conceptual realism and the doctrine of divine simplicity’, Religious Studies, 57(3), pp. 385-402.

Plantinga, A. (2000) Warranted Christian Belief. Oxford: Oxford University Press.

Rohlf, M. (2010) ‘The Ideas of Pure Reason’, in Guyer, P. (ed.) The Cambridge Companion to Kant’s Critique of Pure Reason. Cambridge: Cambridge University Press.

Runzo, J. (1991) ‘Kant on Reason and Justified Belief in God’, in Rossi, P. J. and Wreen, M. (eds.) Kant’s Philosophy of Religion Reconsidered. Indianapolis: Indiana University Press.

Stang, N. F. (2021) ‘Kant’s Transcendental Idealism’, in Zalta, E. N. (ed.) The Stanford Encyclopedia of Philosophy. Metaphysics Research Lab: Stanford University Press.

Strawson, P. F. (1995) The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason. London: Routledge.

Swinburne, R. (2012) ‘Why Hume and Kant were mistaken in rejecting natural theology’, in Buchheim, T., Hermanni, F., Hutter, A., and Schwöbel, C. (eds.) Gottesbeweise als Herausforderung für die moderne Vernunft. Tübingen: Mohr Siebeck.

Van Cleve, J. (1999) Problems from Kant. Oxford: Oxford University Press.

Walsh, W. H. (1939) ‘Kant’s Criticism of Metaphysics: II’, Philosophy, 14(56), pp. 434-448.

Wolterstorff, N. (1998) ‘Is it Possible and Desirable for Theologians to Recover from Kant?’, Modern Theology, 14(1), pp. 1-18.

Leave a Reply

Your email address will not be published.

Follow Us!