DEEP READ · 51
Kritik der reinen Vernunft · Immanuel Kant · 1781／1787
Before Kant, everyone assumed that for knowledge to work, what is in your head must come to match the world outside. Kant turned the arrow around: the mind does not revolve around objects; objects must first pass through our apparatus of cognition before they can show up as objects at all. Suddenly the impossible-looking certainties—mathematics, "every event has a cause"—have a home: they are certain because they are the build rules of the apparatus itself. The price is just as total. The apparatus can only process what appears to us, so every question that reaches past appearance to the world as it is in itself—is the soul immortal, did the universe begin, does God exist—reason is bound to ask, bound to fail to answer, and bound to contradict itself in the attempt. The book is at once a licence issued to human knowledge and the fence drawn around it.
Kant (1724–1804) never left Königsberg in East Prussia (now Kaliningrad), spent most of his working life as a poorly paid unsalaried lecturer, and published this book at fifty-seven after eleven near-silent years. He remarked (paraphrased) that it was the fruit of twelve years' thinking but was written down in some four or five months—which shows: it is hurried and hard, and he conceded it was dry and obscure. Why it is a watershed: European philosophy had split into two camps that refused to concede anything to each other—rationalism (pure reasoning can reach the fundamental structure of reality: Descartes, Leibniz) and empiricism (all knowledge ultimately derives from the senses: Locke, Hume)—and Kant did not pick a side. He ruled that each was half right, then welded the two halves into a single new machine. The 1781 first edition (the "A" edition) baffled nearly everyone; he rewrote large parts for the 1787 "B" edition. Modern texts print both, which is why citations carry A/B page numbers.
In the Prolegomena (1783, his own popular abridgement of this book) Kant wrote the famous line (paraphrased): "I freely admit that it was David Hume's remark that first, many years ago, interrupted my dogmatic slumber." The "dogmatic slumber" was the standard practice of German philosophy at the time: proving things about God, the soul and the world without first auditing what reason is actually equipped to do, as if it could go anywhere it pleased.
Hume's remark was about causation. You watch one billiard ball strike another and the second moves. What you actually see is one thing following another—never the "must," never the hook that welds the two events together. That hook is simply not present in the sensory material. Hume concluded that causation is not a feature of the world but a habit of expectation formed by seeing the same sequence often enough. That cuts to the bone: if causation is mere habit, natural science has no necessity at all.
Kant accepted the first half (the hook really is not in the sensory material) and refused the second (therefore it is only habit). To follow his reformulation you need two distinctions. Analytic vs. synthetic: if the predicate is already contained in the concept of the subject, so that the judgment merely unpacks it, the judgment is analytic ("bachelors are unmarried"—denying it is self-contradictory, but it hands you no new information); if the predicate adds content the subject concept did not contain, it is synthetic. A priori vs. a posteriori: a judgment that needs no appeal to any particular experience is a priori; one you have to go and look to settle is a posteriori. Cross them and you get four boxes:
| A priori: settled without checking | A posteriori: must be checked | |
| Analytic: predicate already in subject | "Bachelors are unmarried"—unshakable but empty; extends nothing | Empty box (unpacking a concept needs no experiment) |
| Synthetic: predicate adds content | 7+5=12; a straight line is the shortest path; every event has a cause—necessary and informative. The whole book exists for this box. | "This stone is warm"—informative, but valid only here and now |
Only the bottom-left box is a miracle: how can anything be independent of experience and still tell you something new?
Kant insists that 7+5=12 is synthetic—a point readers skip past, though it is the doorway to the book. Turn the concepts "7," "5" and "sum" over as long as you like and "12" will not fall out; you have to bring in some intuition (counting through them one at a time in time) to arrive at it. Yet it is plainly a priori. So such judgments exist, and only one question remains: what makes them possible? Kant calls this line of enquiry transcendental, and the word is plainer than it sounds: not a study of objects, but a study of the conditions inside our way of knowing objects that do not come from experience yet make experience possible. (Note: transcendental ≠ transcendent. The first asks, from inside experience, what experience requires; the second reaches past experience toward a beyond—precisely what Kant spent his life opposing.)
The famous analogy in the B preface is the hinge of the book. Copernicus was stuck: assume every heavenly body circles the earth and the numbers refuse to work, with planets forced into backward loops. So he assigned part of the motion to the observer—suppose it is the earth that turns—and the chaos overhead fell into order.
Metaphysics, Kant says, should try the same manoeuvre. The old default was that our knowledge must conform to objects, which makes it permanently mysterious how we could know anything about an object in advance. So reverse it: suppose objects must conform to our way of knowing. Then whatever the apparatus necessarily imposes on any material it processes will necessarily be found in every object it turns out—and that is exactly why a priori knowledge can hold good for all experience.
Here is a less Kantian but very serviceable image. You wear glasses you cannot take off, and the lenses carry a grid. You are entitled to declare, before looking, that everything you will ever see falls on the grid—a claim that is both a priori (no looking required) and about objects (not an empty analytic truth). But you must equally concede that you have no standing to say the things themselves carry a grid. That is the shape of every conclusion in the book: all certainty is bought, and the price is that it holds only on this side of the lens. What really changes is the word "objective": after Kant it no longer means "corresponding to a reality independent of me" but "universally and necessarily valid for every rational subject."
Kant splits the apparatus into two stages: sensibility, which passively receives material, and understanding, which actively organises it with concepts. Take the first stage. His startling claim is that space and time are neither items in the world nor concepts abstracted from experience, but two "forms" sensibility brings to the reception of any material at all. The argument is brisk. You cannot have learned space from experience: to place a sensation "beside me" or "outside that thing" is already to use space—space is the precondition of perception, not its trophy. You can imagine space with nothing in it; you cannot imagine no space, and the same holds for time. They are not the contents poured in, they are the vessel.
From this comes a pair of conclusions that must be held together—drop either half and you misread the entire book. Transcendental ideality: space and time do not belong to things in themselves, only to our mode of reception. Empirical reality: within all experience they are entirely real and absolutely valid. Put concretely: "this table is in space and is 120 cm long" is not an illusion; it is thoroughly objective. It is only that the bedrock of that objectivity is a form of reception we share, not a property of the thing in itself. One asymmetry is usually missed: space is only the form of outer sense, whereas time is the form of inner sense, and every representation—your thoughts, moods and memories included—queues up in time. Time is therefore the deeper of the two: it is the universal format of all mental content.
How does this change the way you see? You can no longer treat "the world I see" and "the world" as the same item—and you cannot slide from that into "then I'm making it all up" either. What you hold is a real, publicly checkable, but transformed report: like a thermometer reading, which is not the heat itself and yet is nothing like an arbitrary number.
That line (A51/B75, paraphrased) is the most quotable and most worth keeping. Concepts with no sensory material are an idling shell—you can talk with great rigour about "the thing in itself" and say nothing whatever; sensory material with no concepts is a smear of pattern in which not even "this is a thing" gets off the ground. Knowledge is always the two working together, and the division of labour cannot be swapped: sensibility can only receive, understanding can only think.
What does understanding organise with? Twelve basic concepts Kant calls the categories—not ordinary concepts distilled from experience like "red" or "dog," but the modes of connection you must employ to bind a manifold of material into "one object." He derives them from the table of judgment forms in traditional logic, in four groups: quantity (unity, plurality, totality), quality (reality, negation, limitation), relation (substance and accident, cause and effect, reciprocity—the group that does the real damage), and modality (possibility, existence, necessity; note that this group adds no content to an object, marking only its relation to our faculty of knowledge—which is the loaded gun for killing the ontological proof of God later on).
Watch "substance" in action. Of the cup in front of you, the senses deliver only a shifting bundle of colour, hardness and glare. "There is an enduring thing whose properties are changing"—the "there is a thing" is not seen; the understanding used the category of substance to tie a heap of changing representations into one bundle. Causation is the same story: Hume was right that you never see the hook, and Kant adds that the understanding hangs it there—and precisely because we hang it, it holds necessarily for every possible experience. Hume's doubt is caught, at the price that causality is now a law of appearances, no longer a law of the world in itself.
Kant said this section cost him more than any other, and it is the hardest thirty pages in the book—but its skeleton is graspable. What has to be shown is: by what right are the categories applied to experience? "Deduction" here is a legal term—producing the credentials for a claim of right—not a mathematical derivation.
The argument starts from something nearly undeniable: "The 'I think' must be able to accompany all my representations." (B131, paraphrased.) Any representation that could not possibly be recognised by me as mine is, for me, as good as non-existent; so all of my experience must be capable of belonging to one and the same "I"—what Kant calls the transcendental unity of apperception, or in plain terms the sheer fact that all my experiences have to hang on one hook. Then the decisive turn: for them to be assignable to the same "I," they must already have been connected according to some rule. A pile of unrelated, randomly flashing fragments does not add up to "one person's experience," any more than random notes add up to a melody you could claim to have heard. And the rules doing that connecting are the categories. Hence: anything capable of becoming "my experience" already obeys the categories, so the categories hold a priori for all possible experience.
The power of the argument lies in its shape—later philosophers call it a transcendental argument: instead of proving something true directly, it shows that if the thing even you concede is to hold, then such-and-such must hold too. It is very hard to shake off, because it counts the sceptic's own footing into the premises: to doubt at all, you need coherent experience first. (The short chapter on schematism that follows patches a seam: how can pure categories, with no sensory colouring whatever, ever meet motley sensory material? By using time as the go-between—substance becomes what persists in time, causality what follows in time by a rule. The pure concept is translated into a time-relation before it can reach the senses.)
Appearance (Erscheinung) = the object as it shows up to us once processed by our forms; the thing in itself (Ding an sich) = whatever it is independently of our mode of reception. Kant's line is that knowledge holds only on the appearance side. Here is the pothole almost everyone hits: the thing in itself is not a truer world hiding behind a curtain. Appearances and things in themselves are not two populations of objects but two ways of considering the same thing. The reason the second concept has to be kept is homely enough: the word "appearance" demands that something be appearing.
But Kant handles it with extreme restraint. The thing in itself is a limiting concept (Grenzbegriff): its entire job is to draw a line at the edge of the map reading "knowledge stops here," not to draw anything beyond the line. You may not say it is one or many, in space or time, caused or uncaused—those words are all categories, and categories work only on the near side. It is a "no entry" sign, not the landscape behind the sign. Both halves of the book's verdict hang on that line: inside it, science finally gets the necessity it had always wanted; outside it, the old questions of metaphysics are ruled unanswerable in principle. For the first time human knowledge has a boundary rather than a to-do list.
The second half of the book, the Transcendental Dialectic, is the best part. "Dialectic" for Kant carries its old pejorative sense: the logic of illusion. First distinguish the third faculty. Sensibility receives, understanding organises experience with categories, and reason (Vernunft) is built to chase the unconditioned—of any explanation it demands "and where did that come from?", refusing to stop short of a terminus that needs no further explaining. This is not a bad habit; it is reason's nature and cannot be shed. But experience contains only one condition after another; the terminus is nowhere in it. So reason unwittingly applies the categories beyond experience and manufactures three ideas—soul, world, God—matching the three branches of traditional metaphysics.
First, the paralogisms—on the soul. Rational psychology argued: since the "I think" is the same subject throughout all thinking, the "I" is a simple, indivisible, persisting substance, and therefore immortal. Kant identifies the switch: the "I" of the "I think" is merely the logical hook on which every representation must hang—a form of thinking. Treating it as an object that has been cognised turns "the unity by which I think" into "a thing I have thought about." You never catch the "I" being observed, because it is not the thing observed but the condition of observing. So the soul's immortality is neither proved nor refuted—it lies outside what knowledge can reach.
Second, the antinomies—on the universe: the same question, with apparently airtight proofs on both sides. Kant lists four:
| Thesis | Antithesis | Kant's verdict | |
| ① The world | Has a beginning in time and limits in space | No beginning, no limits | Both false—"the world as a whole" is not an object that can be given; the question itself is empty |
| ② Composition | Every composite is made of simple parts | Nothing simple exists | |
| ③ Freedom | Besides natural causality there must be a causality of freedom | Everything happens by natural law; there is no freedom | Both may be true—the antithesis holds of appearances, the thesis remains thinkable on the side of things in themselves |
| ④ A necessary being | An absolutely necessary being exists | No such being exists |
The first two are "mathematical" (about composition and magnitude), the last two "dynamical" (about grounds of existence): two diseases, two cures.
In the first two, both sides are wrong, because both share a bad premise: that "the world" lies there as a finished, given totality. On Kant's account the world is not an object at all, only the fact that experience can be pushed further without end—rather like asking where the edge of the earth's surface is. "It has an edge" and "it is edgeless and infinite" are both wrong, because the surface is finite yet unbounded. The question was misframed. In the last two, both sides can be true so long as you say which side you are speaking of. Take freedom, the hinge of Kant's entire philosophy: in the realm of appearance every act of yours obeys natural causality strictly and is in principle fully explicable; and the very same act, considered from the side of things in themselves, may issue from a will not determined by prior causes. No contradiction, because the two claims are not on one level. This does not prove freedom; it proves that the completed determinism of natural science cannot disprove it. A door is left ajar for moral philosophy to walk through.
Third, the ideal—on God. Kant sorts every proof of God's existence ever offered into three: the ontological (from the concept "God = the most perfect being" straight to his existence, since a perfection lacking existence would not be perfect), the cosmological (since something exists, there must be an absolutely necessary being as its ultimate cause), and the physico-theological, i.e. design (the world is so finely ordered that it must have a designer). He argues that the latter two must covertly circle back to the first to say what that necessary being is. And the ontological proof dies on the famous line: "Being is obviously not a real predicate." The example is homely and lethal: the concept of a hundred thalers (the silver coin of the day) and a hundred actual thalers in my pocket are, in conceptual content, not different by a hair. Having them makes me a hundred richer, but it does not add a single property to the hundred thalers themselves. "Exists" adds nothing to a concept; it merely posits that concept, with all its content, in reality. So it cannot be contained in a definition the way "omnipotent" or "all-good" can. (The blow lands squarely on Descartes' Fifth Meditation.)
Then comes a line that often gets dropped: the three ideas yield no knowledge, but they do have a legitimate regulative use—treat them as directions of research rather than results of it. "Proceed as if nature everywhere obeyed unified laws," "as if every organ had a function": such assumptions drive science forward, provided you never swap "as if" for "as a matter of fact." The ideas are a serviceable compass, not a reachable destination. That is also the way to read the most-quoted and most-misread line of the B preface (paraphrased): "I have therefore found it necessary to deny knowledge in order to make room for faith." It is not surrender to religion. By striking God, the soul and freedom off the list of knowable objects, he strikes them off the list of scientifically refutable ones too. The exclusion zone runs in both directions.
Step one, pose the problem. Judgments exist that are both necessary and informative; empiricism cannot account for their necessity (Hume can offer only habit) and rationalism cannot account for their purchase on the real world. Step two, invert. Rather than have cognition chase objects, suppose objects must pass through the forms of cognition to become objects—which locates the source of a priori knowledge: it is those forms.
Step three, dismantle the apparatus and legislate part by part: the Transcendental Aesthetic hangs the necessity of mathematics on space and time as forms of sensibility; the Transcendental Analytic hangs the necessity of the founding principles of natural science on the categories, licensing them by the argument that all experience must be capable of belonging to one "I." Hume's doubt is answered, but by an unexpected route: causation really does not come from the world—it comes from us, and that is exactly why it is necessary. Step four, draw the boundary—all this necessity holds only within the range of processed objects; the certainty of knowledge and the limitedness of knowledge are one fact stated twice. Step five, prosecute the trespassers: reason is built to chase the unconditioned and so must overstep, and what overstepping produces is not wrong answers but unanswerable, self-contradicting questions. The three branches of traditional metaphysics are shut down in one stroke.
So what does the book actually establish? Not that we know more, but that for the first time we know on which stretch of the road we are entitled to say "this must be so," and on which stretch no amount of cleverness buys anything but silence. It is not an edifice of knowledge but a court judgment on reason—in which reason is at once plaintiff, defendant and judge.
Read only the first Critique and you come away with a sceptic who shut metaphysics down—a badly partial picture. This is the first of three Critiques, and the one that clears the ground; the building he actually wanted stands in the other two.
The Critique of Practical Reason (1788) and the Groundwork (1785). The first Critique asks what I can know; here the question is what I ought to do. His starting point: if a moral law is to bind at all, it cannot rest on anything empirical—not divine reward and punishment (that is prudence), not fellow-feeling (here today, gone tomorrow), not calculation of consequences (which never finishes). It can only be a law reason gives itself. Hence the categorical imperative—"categorical" meaning unconditional, as against a hypothetical imperative ("if you want to be healthy, exercise," which lapses the moment the condition does). The two most usable formulations: "Act only on that maxim which you can at the same time will to become a universal law"—tested by imagining everyone doing it, so that borrowing with no intention to repay, universalised, leaves nobody willing to lend and cancels itself; and "Always treat persons, yourself included, at the same time as an end, never merely as a means"—note the "merely": using a taxi driver's service is fine so long as you have not reduced them to an instrument. All of it stands on the gap the first Critique left open: freedom cannot be known, but the reality of the moral law vouches for it in reverse—you ought, therefore you can.
The Critique of Judgment (1790) patches the seam between the other two: on one side appearances fixed by natural law, on the other the domain of freedom—how do they share a world? His bridge rests on two things. Beauty: an aesthetic judgment is a "disinterested pleasure"—finding a flower beautiful has nothing to do with wanting to own it or with what it is good for, and yet the judgment "demands universal assent without a concept." There is also the sublime: before a storm-driven sea or a limitless sky you are first overwhelmed and then oddly exhilarated, because you notice that nature can crush your body but not the reason in you that can conceive the infinite. Teleology: living things look built "for" something, though science forbids explaining by purpose; his solution is that "as if purposive" is a stance we cannot avoid taking, not a property nature possesses.
Two short essays are his most readable. "What Is Enlightenment?" (1784) defines it in a line: "Enlightenment is the human being's emergence from self-incurred immaturity"—immaturity being the inability to use your own understanding without another's guidance, "self-incurred" because the cause lies not in intelligence but in want of courage. Hence Sapere aude: dare to know. "Toward Perpetual Peace" (1795) sketches a startlingly modern programme—republican constitutions, a federation of free states, a universal right of hospitality—widely regarded as one source of the League of Nations and of later democratic-peace theory. Add all this back and you see the whole Kant: he fenced off knowledge not to silence anyone, but to clear ground on which "human beings are free, and therefore have dignity" could stand.
1. The book asks one question: how can there be knowledge that owes nothing to experience yet still tells you something new? (7+5=12; every event has a cause.) Hume said such necessity is only habit; Kant took that to be a death sentence on the necessity of science.
2. The answer is a Copernican inversion: cognition does not conform to objects; objects must conform to the forms of cognition in order to be objects at all. You wear glasses you cannot remove whose lenses carry a grid—so you may declare in advance that all you see falls on the grid, but never that the things themselves carry one.
3. Space and time are not containers in the world but the two forms in which we receive all material. Hence the paired verdict: transcendentally ideal (not properties of things in themselves) and empirically real (utterly valid within experience). This is not "it's all illusion"—it is where objectivity comes from.
4. "Thoughts without content are empty; intuitions without concepts are blind." Concepts alone idle; sensation alone is a smear of pattern. Sensibility can only receive, understanding only think, and the roles cannot be swapped.
5. The twelve categories are not distilled from experience; they are the connections the understanding must use to tie a manifold into one object. Hume was right that you never see the hook of causation—the understanding hangs it there, and precisely because we hang it, it holds necessarily for every possible experience.
6. The Deduction rests on one sentence: "The 'I think' must be able to accompany all my representations." Unrelated fragments no more add up to one person's experience than random notes to a melody; to belong to one "I" they must already be rule-connected—and those rules are the categories.
7. The thing in itself is not a truer world behind a curtain but a limiting concept: its whole job is to draw a line at the map's edge reading "knowledge stops here," and to draw nothing beyond it. You cannot even say it has a cause—that would already be using a category.
8. Reason is built to ask "and where did that come from," and so must overstep: mistaking the formal "I think" for a soul-substance; treating the never-given totality of the world as an object; and the proofs of God—the ontological one dies on "being is not a real predicate": the concept of a hundred thalers and a hundred thalers in your pocket differ by nothing in conceptual content.
9. The antinomies get two verdicts: in the first two, both sides are false, because "the world as a whole" is not an object that can be given and the question was misframed; in the last two, both can be true—the complete determinism of appearances cannot disprove freedom, because they do not operate on the same level. That gap is the door left open for moral philosophy.
10. "Denying knowledge to make room for faith" is not surrender but a two-way closure: God, soul and freedom can neither be proved nor refuted. What the book finally establishes is not that we know more, but that for the first time we know where we are entitled to say "this must be so," and where no amount of cleverness buys anything but silence.