How Hume woke Kant from his ‘dogmatic slumber’

Tyler asked

How does Hume’s extreme skepticism influence the thinking of Immanuel Kant? How does Kant resolve the perceived dilemma introduced by Hume’s scepticism?

Answer by Graham Hackett

One year ago I would have answered this question in a quite different way to how I am going to (try to) answer it here Last year, I would have been rather more dogmatic about both Kant and Hume than I am now. I am less sure. I ask you to bear in mind this, when you read my response to your question. It is not that I am trying to show Socratic humility; it is just that the degree of subtlety in both of these writers is immense, and I don’t think it is possible ever to be definitive.

Kant was very impressed by Hume, and remarked that he had been ‘woken up from a long slumber’ after reading him. Hume was doubtful about how much we could know through reason, and regarded empirical matters of fact, ideas and impressions as being all important. Added to this was the stunning success of the scientific model of knowledge; just a few laws developed by Copernicus, Kepler, Galileo and Newton enabled the development of large and impressive bodies of knowledge. It is no wonder that Hume took a very dim view indeed of metaphysics, and dismissively opined about it in the following words;

“If we take in our hand any volume of divinity or school metaphysics, for instance, let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames: for it can contain nothing but sophistry and illusion.”

I do not think that Hume was an extreme sceptic in the normal sense of the term; viz, that we can never know anything. This reputation partly comes from his view on the respective roles of empirical knowledge and reason. For Hume reason was involved in considering the relations between ideas, and was either a priori reasoning or based on matters of empirical fact. For some 18th century philosophers, like Clarke, reason was a much weightier matter, with metaphysical implications. For instance, Clarke, as a moral realist, thought that reason extended to matters of ethics as well as matter, for our actions would be fit or unfit even if no one could have any intuition that they were so. Physical matter and the fitnesses or unfitnesses of actions exist independently of us and are there for reason to discover. Hume, however, did not believe we have access to anything but ideas and impressions, so he thought that Clarke’s view must be incorrect; we cannot grasp anything outside the bounds of experience.

Hume also has critical views concerning causality and determination. He sees no causal laws in operation; only the constant conjunction of events A and B, from which we infer that A causes B.

Kant was enormously disturbed by the success of the scientific revolution and by Hume’s scepticism about such matters as the role of reason, causality, the timeless underpinning for our ethical beliefs and the usefulness of metaphysics in general. He wished to rescue metaphysics from the demise that Hume predicted for it because of the success of science. It is no coincidence that he referred to his book Critique of Practical Reason as beginning a ‘Copernican revolution’ for metaphysics, putting it on the same firm foundation as science.

It is possible to read into Kant’s division of existence into two worlds — noumena and phenomena — as a response to the attacks of the empiricists on metaphysics. The world of phenomena is the world which can be known empirically, by scientific discovery. It is the world governed by the natural law, and everything in it is structured by time, space and causality. Because we are part of this world, we are also governed by the natural law and our behaviour is determined. For Hume this is all there is.

But for Kant, there is also the noumenal world, which is outside space and time and causality via the laws of nature. We can know nothing certain about what is in the noumenal sphere, but because of Kant’s adoption of the ‘two standpoints’ we are part of both worlds. From the standpoint of theoretical reason, human actions are phenomenal events occurring in the natural world and are therefore completely determined by natural physical laws. However, from the standpoint of practical reason, actions are noumenal events that result from a free will that deliberates between alternatives, evaluates them, selects one, and thus acts freely by self determination. So humans are determined when viewed (theoretically) from a third-person perspective as an object, but free when viewed (practically) by the ‘self’ from a first-person perspective as a subject. This is a rather clever way of allowing us to have free will and yet still recognise the necessity of the laws of nature. It is also a good way of showing that despite the successes of science and empirical method, we can still find a powerful role for reason. According to Kant we can even formulate synthetic a priori knowledge, where reason is used, independently of experience to know things which are not self evident.

Hume would probably not have been convinced by Kant’s defense of reason and freedom against empiricism and the natural laws, but he would certainly have been impressed.


Answer by Craig Skinner

Hume’s views greatly influenced Kant.

Early in Prolegomena to any Future Metaphysic (1785; trans. Bennett J: online at, page 2) Kant says:

“David Hume’s attack on metaphysics was more decisive for its fate than any other event… since the earliest recorded beginnings of metaphysics’, and ‘It was my recollection of David Hume that broke into my dogmatic slumber.”

Kant felt he had cracked the Hume problem. Later in this text (page 36) he says:

“So the Humean problem is completely solved, though in a way that would have surprised its inventor… the complete reverse of anything that Hume envisaged — instead of the concepts (of the understanding) being derived from experience, that experience is derived from them.”

What then was the problem posed by Hume, what is Kant’s solution. and is it a good one?

Hume problem. Hume held that all knowledge falls into one or other of 2 categories (a view later termed ‘Hume’s Fork’ or ‘Hume’s Dichotomy’, and I take it this what your ‘dilemma’ refers to):

* matters of fact
* relations of ideas

Matters of fact are known from experience (known a posteriori), tell us something about the world, and are contingent truths (could have been otherwise) e.g. Paris is the capital of France. Relations of ideas are known simply by grasping the meaning of the ideas (known a priori), are necessary truths (couldn’t have been otherwise), but tell us nothing about the world e.g. truths of logic. Any statement which is neither a matter of fact nor a matter of logic can’t be knowledge. Talk of God’s essence and actions, immortality of the soul and other metaphysical ideas, are hot air. Hume’s famous last paragraph in his ‘Enquiry Concerning Human Understanding’ (1748) reads:

“If we take in our hand any volume; of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matters of fact and existence? No. Commit it then to the flames; for it can contain nothing but sophistry and illusion.”

So, for Hume, metaphysics is bunk.

Furthermore, he felt that the role of human reason was overblown. Reason is only the servant of our feelings, helping us to plan the means to the ends set by our feelings, was his view. We think reason tells us there is an external world, an enduring self, and a necessary cause-and-effect relation, whereas we don’t really know there is an external world, we don’t actually see any necessary causal connection (only constant conjunction), and introspection reveals only a bundle of sensations and thoughts, no enduring ‘I’.

This extreme scepticism was too much for Kant. He wanted to show that metaphysics is possible (and indeed to write some actual metaphysics), and that talk of external world, causation and self was not empty. And that reason had a bigger role.

Kant’s solution. Kant agreed that if Hume’s Fork truly were an exhaustive account of kinds of knowledge, then metaphysics would indeed be impossible, for experience can’t justify a world, causation and a self; and purely analytic a priori reasoning does just yield what is already implicit in the concept we start with. So Kant cleverly suggested that Hume had overlooked a third type of knowledge, a third prong on the fork, as it were, one on which metaphysics could hang. He suggested that in addition to matters of fact (synthetic a posteriori knowledge) and matters of logic (analytic a priori and necessary knowledge) there was synthetic a priori knowledge i.e. necessary truths, known a priori, but which, unlike analytic truths, did tell us something about the world. In short, synthetic a priori knowledge makes metaphysics possible. But is there such a thing as synthetic a priori knowledge? Kant gives examples.

First, mathematics. Kant maintains that ‘7+5=12’ is not an analytic truth, known just by understanding the meaning of the numbers. The concept ‘7+5’ contains the uniting of 7 and 5 into a single number but doesn’t contain 12. We obtain 12 by amplifying the concept, using (at first) say our fingers to count on. This is easier to grasp with bigger numbers, say the concept ‘38976+45204’ which clearly doesn’t contain ‘84180’ So, whilst 7+5=12 is known a priori, it is not analytic, it clearly tells us something about the world and so is synthetic. Next, physics. Kant maintains Newton’s laws are known a priori yet apply to the world, and do so necessarily. Another example of synthetic a priori knowledge. Coming to metaphysics, Kant says the concepts of space, time, enduring objects moving in space/time and interacting causally, are all known a priori because these concepts are necessary for any rational mind to experience any kind of coherent world. In short they are the preconditions for any experience at all. Space and time are the forms of our sensibility (perception) and things with properties (substances and accidents to use the old terms) and causality are categories of our understanding, to use Kant’s technical terms. We don’t get these concepts by experiencing the world (as Hume thought), we are only able to experience any world at all by organizing our sensory input and thoughts according to these concepts. And so, the world, the self and causation are all restored. But the price of this, says Kant, is that we can only ever know what we experience, how things appear, never how things are in themselves, about which we can know nothing.

A good solution? We get back a world, a self, and causality, but we know these only of the world as experienced, not the world in itself. It’s a brilliant and novel tour-de-force of fancy philosophical footwork. But does it take us any further? One could imagine Hume saying, fine, you’ve explained why the world (as experienced) shows causation, but only because you put it in as a category (of our understanding), whereas I say we take it from the world by experiencing constant conjunctions; you derive it a priori and say it’s necessary, I derive it a posteriori but say no necessity can be seen, although of course we can’t do without the notion of causality both in science and in everyday life. But at least I consider that my constant conjunctions apply to the real world, whereas your necessary concepts and forms only apply to appearances. Scholars still argue the matter.

And this assumes that synthetic a priori knowledge exists. Most philosophers think mathematical truths are analytic, although attempts to reduce maths entirely to logic (Frege, Russell, Whitehead) have not succeeded. However Newton’s laws of gravity, far from being necessarily true, are not even true. But I forgive that. Pretty well all 17th and 18th century philosophers, and other thinkers, including Kant and Hume, revered Newton, thinking his laws to be the last word, and Hume models his intended science of the mind on Newton’s science of matter.

I wish I could get Kant and Hume into a room, give each of them a copy of Darwin’s ‘Origin of Species’ (‘aha, of course’, I might overhear) to read and digest, and internet access to neuroscientific findings, And then to offer the following paragraph for their comment:

Our cognitive capacities are part of our evolved nature. They are likely to be adapted to important features of the way the world is, thereby favouring our survival. Those among ancestral human populations who had less well-fitting capacities left fewer descendants. We would expect that enduring aspects of the world might be hard-wired into our brains so that each generation doesn’t have to start from scratch. Features like space, time, objects and causal interaction. And indeed, experiments with very young babies indicate that they have the notions of space and of causation. Being hard-wired, this knowledge is a priori, and is necessary because all creatures necessarily evolved in such a world. And so Kant is right here. On the other hand, this account suggests that these are features of the real world, the world-in-itself, not just of the world of appearances. So Kant is not so right there. Furthermore, the hard-wiring is the end result of the cumulative experience of our ancestors (improved survival/reproduction in those who had favourable prototypes of our present cognitive capacities). So, ultimately, all knowledge is from experience, either our own or that of generations of our forebears, and so a posteriori. There is no a priori knowledge at all (empiricists rejoice), so it’s fruitless to worry about whether some of it is analytic and some synthetic.

Would they come to an agreement? From what I know of great philosophers, they might agree partially or on details but not completely on such a major issue.

Forgive my broad-brush, non-expert, punter’s account of (some of) Kant’s views. Kant is very difficult, systematic, subtle, wordy, sometimes obscure, even inconsistent. A lifetime could be spent trying to fully grasp him. And has been by some, but not by me. Incidentally, the short ‘Prolegomena to any Future Metaphysic’ is the best introduction to Kant’s writing on metaphysics, so clear, snappy and vivid, it’s difficult to believe it’s written by the same man who, four years earlier, wrote the long, dense, detailed, sometimes obscure and tedious ‘Critique of Pure Reason’


3 thoughts on “How Hume woke Kant from his ‘dogmatic slumber’

  1. Does kant truly state that categories of space, time and causality apply to the real world, the world in-itself? Or does he say that these synthetic apriori concepts are only applicable to the phenomenal world, the world of appearances?
    You state that kant would be wrong to commit himself to the former, but does he claim that at all? I believe that he posited the latter instead. How could he admit that this synthetic a priori knowledge applies to the noumenal world if it is even derived from experience, but rather understands it as a necessary structure of the mind that helps us make sense of the phenomenal world.
    This question refers to your revision of both hume and kants arguments with relation to darwinian knowledge that we now posses.
    Thanks for your reply.

  2. The problem with your ‘evolutionary’ argument, is that if it’s true, then it undermines reason, and if it undermines reason, then how can your argument be reasonable? You’re just one h. Sapiens appealing to another h. Sapiens, both of whom labour under the same delusion.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.