What was Kant up to in the first Critique?

What was Kant trying to do in the first Critique? Why did he propose the apparatus of the categories that supposedly shape all thought and experience? The central problem of the Critique of Pure Reason is to understand a priori knowledge and especially what he called synthetic a priori knowledge, which is something we know apart from experience, but yet which is not contradictory to deny. How does this fit together with the categories and the forms of intuition?

Let me try to fill in a little background. Kant is not a skeptic. He doesn’t doubt that we know things or that we know approximately what we take ourselves to know. He has argued that we do have a priori knowledge or at least what we take to be a priori knowledge. (See the Second Edition Introduction, B3 - B6.) In the sense he’s interested in, practically everyone except the most extreme skeptics agrees that we do have a priori knowledge. (If you think that you don’t need to go look to find out if there are round squares on Venus or whether twice two still equals four on Mars, then you believe in a priori knowledge.)

In addition, to understand what he’s up to, you need to understand the analytic-synthetic distinction, which he made famous. (Other people had made the same distinction before, but the name for it is his.) Now, I don’t think a sharp analytic-synthetic distinction will stand up, basically for Quinean reasons, but one can’t understand what Kant was up to (or a lot of other things in philosophy) unless you have some sense of the distinction and why it seemed plausible.

Here’s a bare-bones version: An analytic judgment is one that, if true, is true in virtue of its meaning. (This isn’t quite Kant’s way of putting it, but it will do for now.) An example would be “all bachelors are unmarried.� You don’t need to go out and conduct social science polls or examine marriage records at the courthouse to find out if that’s true. All you have to do is understand what the terms mean – in Kant’s terms, know what’s included in the concepts – to see that all bachelors must be unmarried. When you’ve got a true analytic judgment in hand, all you have to do is analyze it to find out that it’s true.

What, then, is a synthetic judgment? One that, if true, is not true solely in virtue of its meaning. For example, “all bachelors are over four feet tall.� If that’s true, it’s not true in virtue of the meaning of the constituent terms. To find out if it’s true, you’d have to actually examine bachelors and measure their heights. A single four-foot bachelor would show that it’s not true that all bachelors are over four feet tall, but you’ll never find out whether there’s a four-foot bachelor somewhere without checking. Analyzing your concept won’t give you the answer. Instead, you have to synthesize or bring together in a single judgment the concept of bachelor with the observed (or measured or reported or ...) facts about the heights bachelors actually have.

When you put it this way, the distinction can look inevitable: For judgments that are capable of being true or false, if they are true, then either they are true in virtue of their meaning ... or else they’re not. If they are, they’re analytic; if not, they’re synthetic. I will not here try to explain why I think the distinction nevertheless breaks down. My only point, preparatory to some further discussion of Kant, has been to indicate why it seemed plausible in the first place.

Let’s go back now to the a priori (and the a posteriori). This is a distinction in how we know or come to know things. Something that can be known a priori can be known in a way that is in some sense prior to or apart from experience. (Experience might still be necessary to form the concepts in terms of which we can know something a priori. E.g., experience is needed to form the concept of bachelor, but no further experience is needed to know that all bachelors are unmarried.) In contrast, something is known a posteriori if it can only be known on the basis of experience.

If you put this together with the analytic-synthetic distinction, you get a four-way classification:

Kant thought there are clear examples for box 1, the analytic a priori, like “all bachelors are unmarried.� Likewise, there are clear examples for box 4, the synthetic a posteriori, like “some bachelors are over four feet tall.� He thought box 2 was empty (perhaps a bit too hastily) but that the interesting case was box 3. Were there any examples of a priori knowledge that are not analytic?

He thought it clear that there were. In common sense, in mathematics, and in natural science, we find people appealing to and relying upon principles that are not analytic, but also couldn’t be learned from experience. If we know them at all, we know them a priori. Some of his examples are that every event has a cause, that a straight line is the shortest distance between two points, that two straight lines can’t enclose a space, and that things don’t pop into or out of existence (he called this “the permanence of substance,� but he was thinking of the conservation laws of physics). You may or may not find these examples convincing, but they are the kind of things he was thinking of (and he is surely right that none of them is analytic).

[There’s an interesting argument that Kant didn’t use (it’s from Ewing, I think) that tries to undermine objections to there being some things we know that are both a priori and synthetic that goes like this: It isn’t analytic that there are no a priori synthetic truths; that is, it’s not self-contradictory to assert that there are some. So, if there are no synthetic a priori truths, that’s a synthetic truth. Is it also an a posteriori truth, i.e., one we learn from experience? If so, then what the opponent of the synthetic a priori should say is that he hasn’t so far found any examples he finds convincing, but one might turn up. What the opponents of the synthetic a priori usually do say, though is that there can’t be any such truths. But they could only know that, if at all, a priori. So, if they think they know it, their position is self-referentially inconsistent. They would be believing that it’s a synthetic a priori truth that there are no synthetic a priori truths.]

His basic argument for its being possible for us to know some things that are synthetic and a priori is that we actually do. “Since [pure mathematics and pure natural science] are actually given, it is surely proper for us to ask how they are possible; for that they must be possible is proved by their being actual.� (B 20-21) Since he’s not a skeptic, he doesn’t take seriously the suggestion that we’re just completely wrong across broad stretches of our intellectual life. You might view him as posing the rhetorical question, If this isn’t knowledge, what is?

To continue, after having argued that there is knowledge of synthetic a priori truths, Kant asks how it is possible for there to be. You might wonder why pursue that question – if we really do know there are synthetic a priori truths, which therefore must be possible. I think at least four reasons might be given. First, we may just find the question interesting. The marks of the a priori, according to Kant, are necessity and strict universality (i.e., universality that does not admit of even the possibility of an exception). But the kind of necessity we understand best is logical necessity – the necessity of that which is self-contradictory to deny. There’s something puzzling in the idea that there is a kind of necessity that we are capable of grasping apart from experience but which is not logical necessity. Second, an explanation of how it is possible for us to know synthetic a priori truths might provide an answer to various skeptical doubts (such as Hume’s about causality, since Hume’s argument was that causal relations were neither analytic nor guaranteed by experience – so they could be answered if we know of causality in some way that is neither analytic nor guaranteed by experience, i.e., if our knowledge of it is synthetic a priori). Third, figuring out how it is possible may tell us something about the range to which it is applicable. Fourth (which may be an application of the third), many of the traditional questions and arguments associated with metaphysics (is there a God? a soul? freedom of will?) seem to involve appeals to principles that, if we know them at all, must be synthetic and a priori. If we understood how it was possible for there to be a synthetic a priori, it could shed light on those debates.

Kant’s basic picture of experience is that, in the full-blooded sense, it involves the coordinated contribution of both the senses and our conceptual capacities. There is ‘input’ from the senses – which he calls ‘intuition’ – that is given to us by the world, and there is what we do with that input as we cognitively engage it, classify it, and make judgments. This is what he’s talking about when he says: “The form of experience is nothing but the synthetical unity of phenomena according to concepts.�

About each of these factors, Kant argues that we can distinguish the form it takes from its particular content. Let’s start with intuition: He says we can imagine many different kinds of content for sensory input, but one thing that we can’t even imagine being different is its being presented in space. To imagine it not being in space is just not to imagine anything at all. We can even, he says, conceive an empty space – “think away� any sensory input at all – but we can’t think away space itself. Similar remarks apply to the ordering of our inner experience in time. Thoughts, sensations, feelings, emotions, mental images and so forth are always experienced as having duration and being in a temporal order. No matter how different they might be, we cannot think away their temporality. Again, to do so would be just to imagine not having any inner experience rather than imagining some different kind of inner experience.

Thus, space and time are forms of intuition. Though all our experience, inner and outer, is ordered in space and time, we do not acquire the concepts of space and time from experience, because, no matter how different our experience was, it couldn’t be experience at all (for beings like us) apart from the ordering in space and time.

In a similar way, Kant holds that in thinking about the world, making judgments about how it is, there are basic concepts (he calls them ‘categories’) in terms of which we must do so. We think, for example, in terms of causality and substance, which we could not learn from experience but cannot avoid applying to experience.

And it is as the product of these two distinguishable factors in our experience that Kant explains our capacity for synthetic a priori knowledge. We can think of “things in themselves,� that is, things considered entirely apart from any cognitive engagement on our part. We can also think of “appearances� or things as we are cognitively engaged with them. These are the same things – Kant does not have a two-worlds view! About appearances – which covers everything that shows up in experience, we can say that they must conform to the forms of space and time and to the categories, not because these are necessary features for anything to exist, but because they are necessary for us to experience them. Anything that didn’t so conform couldn’t show up in our experience. The forms of intuition and categories are like a filter or grid that we impose upon the world. As for things, considered as they are in themselves, all we can say is that they exist (they are, after all, what appears in experience), but beyond that we can justify no synthetic claims about them.

Now, if this is right – in particular, if synthetic a priori knowledge is possible only insofar as it applies to the world of possible experience^§ – that is, experience that is possible for beings like ourselves – then we have to admit that what goes entirely beyond possible experience – including, as mentioned above, many traditional arguments in metaphysics bearing on the existence of God or the immortality of the soul – are not and cannot be matters of knowledge. All of the arguments fall down somewhere by assuming that principles we know to apply within experience also extend to things considered in themselves. As Kant said, “Thoughts without content are empty; intuitions without concepts are blind.â€? (B 75) Without the sensory input, we have only concepts spinning in the void; without conceptual articulation, there is only a sensory chaos. In a striking metaphor, he said, “When the light dove parts the air in free flight and feels the air’s resistance, it might come to think that it would do much better still in space devoid of air.â€? (B 8-9) We need the atmosphere of experiential input for the wings of thought to beat against and only deceive ourselves if we imagine we can do without it.