Kant: An Overview  

A.  Kant's Motivations

One of the things that make Kant's work so difficult is that he was motivated by so many different concerns. But for our purposes we may emphasize two. There was, first of all, an epistemological motivation: a desire he shared with Berkeley to thwart the apparently skeptical implications of Descartes' and Locke's assumptions. (In this Kant was able to profit from the writings of David Hume, who brilliantly, explicitly, and thoroughly drew the skeptical conclusions that Locke had never worried much about and Berkeley had struggled to avoid.)

At first, Kant's way of posing the skeptical problem may be puzzling. The central question of his great, primarily epistemological work the Critique of Pure Reason is: How is synthetic knowledge a priori possible? By this Kant means to ask how it is possible for us to acquire knowledge about the world without needing to infer this knowledge from experience. ("Synthetic" means "not analytic," and a judgment is analytic if it is true not because of the way the world is but simply in virtue of the meanings of the terms or concepts it contains, as "Cats are cats" and "All bachelors are unmarried" are true. "A priori" is contrasted with "a posteriori." An a posteriori truth is a truth known on the basis of experience; a truth is a priori if it can be known "prior to" or without the need of experience.) Kant's claim that we can know facts about the empirical world a priori is striking, but as a response to skepticism it may seem a bit odd; it seems that showing the possibility of a posteriori knowledge would be enough to thwart the skeptic without the need of such a dramatic claim as Kant wants to make.

To understand why Kant sees a defense of synthetic a priori knowledge as necessary to defeat skepticism, we need to glance briefly at Hume (who, Kant famously remarks, awoke him from his "dogmatic slumber"). Hume held that all knowledge concerned either relations of ideas or matters of fact. This is, more or less, Kant's distinction between analytic and synthetic truths. Hume held further that knowledge of the relations of ideas could be attained by intuition and demonstration -- that is, in Kant's terminology, could be attained a priori. So far Kant and Hume agree. But Hume goes on to claim that all our knowledge of matters of fact comes from experience -- i.e.., more or less, that all synthetic knowledge is a posteriori. How do we arrive at knowledge on the basis of experience? There are two main ways. First of all, there is observation: if we observe something we can know of its existence (though, as we remember from Berkeley, there are problems here about just what we can properly be said to observe). (Sometimes Hume includes the memory of past observations as a separate source of knowledge.) But clearly very few of our beliefs about the world are beliefs about things we have directly observed. The vast bulk of what we think of as our knowledge requires Hume's second main method of acquiring knowledge: experimental inference. From our observations, we make inferences about what must cause or explain them. Hume suggests that experimental inference always has to do with cause and effect. (We might prefer the claim that it always has to do with explanation, a more general notion, but this will not affect Hume's arguments.) For example, from our observation of footprints we conclude that someone has walked this way. But of course to arrive at this conclusion we need more than just our observation of the footprints: we also need knowledge of the causal laws relevant to the production of footprints.

So far the conclusion is that most of our synthetic knowledge comes from experimental inference, and experimental inference requires knowledge of causal laws. Most of our synthetic knowledge, therefore, requires that we already know causal laws. Hume therefore investigates the possible sources of this knowledge. The experience which leads us to believe that A-type events cause B-type events, Hume maintains, is the experience of constant conjunction: that is, if we observe that stepping on soft soil is regularly followed by the production of footprints, we may conclude that stepping on soft soil causes footprints. But the statement that stepping on soft soil causes footprints implies that the first is always followed by the second (or at least that there is some more refined statement of a causal law which is genuinely universal), and this in turn implies a prediction about the future, namely that stepping in soft soil will continue to be followed by footprints. How can we know that this prediction is true? In general we know that something is true only if we see that it is true (i.e. if we observe it, intuit it, or remember it) or if we infer that it is true from what we observe or intuit by either deduction or experimental inference. We can't see into the future, which leaves deduction and experimental inference. But deduction clearly will not work, since there is no contradiction involved in the supposition that the future will not resemble the past. It does not seem to be analytic, a matter merely of the relations of ideas, that the future will be like the past. But this leaves only experimental inference, which Hume has already argued always depends on knowledge of cause and effect. But we have now completed a circle; we were looking for a justification of our knowledge of cause and effect, and seem to have found that it depends on knowledge of cause and effect, which is no help.

Worse still, we seem to have arrived at a very general skeptical conclusion. Most of our knowledge depends on experimental inference; experimental inference depends on knowledge of cause and effect; but there is no way to acquire such knowledge. It looks as though experimental inference does not really give us knowledge at all, and thus that we in fact know very little about the world: just what we can directly observe or remember, and this is very little indeed (even apart from the further skeptical worries that can be directed at observation and memory).
Now, Kant agrees with Hume that most of our knowledge depends on knowledge of causality, and agrees further that we cannot acquire knowledge of e.g. the law of causality a posteriori. But Kant hopes to be able to show that we can have knowledge of causality a priori; since knowledge of causality is necessary for most of the rest of our knowledge, Kant's defense of synthetic a priori knowledge is by implication also a defense of most of our ordinary a posteriori knowledge.

So much, then, for Kant's epistemological motivation, the desire to defend knowledge of the natural world from the Humean skeptical challenge. (Really this is not Kant's only epistemological motivation; there is also the desire to explain how it is possible for mathematics to be known a priori and yet to be so useful in thinking about the empirical world; and also to explain why the progress made in philosophy had been so miserable while mathematics and natural science had made tremendous advances in knowledge.) Kant also has a practical motivation. If his epistemological motivation is to protect natural science from the dangers of Humean skepticism, his practical motivation is to protect human freedom from the encroachments of natural science. The Newtonian world-view Kant wants to defend, according to which the universe is a kind of giant machine, any event in which has its own cause, seems to squeeze out the possibility of any genuine human freedom; it may leave room for feelings of freedom, but if every human action, like other natural events, is entirely the result of prior events, it is hard to see how such feelings of freedom can be more than a sham. Kant wanted not only to defend the Newtonian picture of the universe, but also to show that this picture still left room for meaningful human freedom.
This conflict between freedom and determinism arises in various ways for various thinkers, and we can be a bit more specific about the way it arises in Kant's thought. Kant finds it a particularly pressing issue because of the relation between morality and freedom. Of course, virtually everyone who has thought about morality thinks that there is a close link between morality and freedom: it does not make sense to hold someone responsible for an action if the agent was not free not to do it, so moral evaluation only makes sense to the extent that the agents being evaluated are free. But for Kant the relation between freedom and morality is particularly close: Kant's conceptions of freedom and of morality lead to the view that acting freely and doing the right thing are in the end identical.

The connections here are numerous and complex; I will mention two. First, a nice element of Kant's picture is this: he holds that the moral law comes from within each individual. It is the same for everyone, not because it is imposed on everyone from the outside, but just because everyone's reason shares certain features. Kant sees the requirements of morality as formal features of reason (in something like the same way he sees space and time as "forms of intuition": just as everyone perceives the world as spatiotemporal, everyone recognizes the same moral commands). But because the moral law comes from within individuals, moral action does not require external demands or constraints. Morality is not something forced on someone from the outside. You don't act morally because the law tells you to, or your parents tell you to, or God tells you to, but because you know you should. So moral action is free rather than compelled, "autonomous" rather than "heteronomous."

This suggests that whenever one acts morally, one is thereby acting freely. The converse, that acting freely is always acting morally, is much more controversial, and Kant's reasons for it have to do with features of his thought that I have not yet discussed. But I can at least hint at one of them. Kant seems to have three different accounts of what freedom is, but they are three accounts which he believes are very closely related. The three conditions Kant thinks are required for freedom are: (1) that my actions are brought about by reason rather than by impulses, desires, or urges; (2) that I originate my actions (I am a "first beginning") rather than just being a middle link (a "subordinate beginning"), i.e. if the cause of my actions is not itself caused; (3) that I do things not because of what is but because of what ought to be. If I understand Kant's view correctly, he thinks that (1) and (3) are each both necessary and sufficient for freedom, while (2) is necessary but not sufficient. 

These three conditions are linked by two theses. The first condition is linked with the third by way of the thesis that reason is concerned with what ought to be the case rather than with what is the case. At work here is Kant's threefold division of mental operations into what he calls "faculties": sensibility, understanding, and reason. "Sensibility" comes from "sense," and has to do with our sensory experience of the world; "understanding" is concerned with classifying our sensory experience and making judgments about it. Sensibility and understanding combine to give us knowledge about the empirical world, about how things actually are. But while the function of sensibility and understanding, working together, is the theoretical one of providing us with knowledge, the function of reason is the practical one of providing us ideals that we should strive for, of providing us with a picture of the perfect society, for example, in order that we can guide our actions toward something genuinely worthwhile rather than merely reacting to society as it happens to have been organized so far through a more or less arbitrary history of greed and chance and struggles for power. (Similarly, reason also provides us with an ideal of explanation that we ought to strive for in the sciences, whether or not this ideal is actually attainable.) Thus actions guided by reason will be actions directed at achieving what ought to be the case, not mere reactions to what actually is, and satisfaction of the first condition guarantees satisfaction of the third.

This helps us to see why Kant thinks that genuinely free action is necessarily moral. The link between the first condition and the second, between acting on the basis of reason and being a "first beginning," is a crucial part of Kant's attempt to reconcile determinism and freedom, and we will come back to it; for now I will only note that on Kant's view the notions of time and causality are simply not applicable to reason.

We now have a reasonably good view of those two of Kant's motivations with which we will be most concerned: his goal of defending Newtonian science from Humean skeptical challenges, and his goal of defending freedom from the encroachments of Newtonian science. It is, in part, the difficulty of accomplishing both these ends at once that accounts for the characteristic complexity and difficulty of Kant's work.

B. Kant's Epistemology

Kant's attempt to satisfy both of his motivations centers around the distinctive view he termed "transcendental idealism." To understand Kant's transcendental idealism it is useful to begin by returning to his description of our mental faculties. (In this paragraph and the next four I am drawing on Lewis White Beck, "Kant's Strategy," in his Essays on Kant and Hume.) Kant made a remarkable attempt to synthesize the epistemological views of the rationalists and the empiricists, empiricism being represented for him primarily by Hume and rationalism primarily by Leibniz and his German disciple Christian Wolff.

As Kant saw it, each tradition recognized only one source of knowledge, and correspondingly each thought of all mental contents on a single model. We have already seen the empiricist tendency to think of all mental contents on the model of visual impressions; as Hume put it, ideas were just faint copies of sense-impressions. And we have already seen how this view contributes to skepticism: if all our evidence about the way things are comes from our perception of sense-impressions, then it seems that all we can know about with any certainty are the sense-impressions themselves. And we have seen in particular Hume's skepticism about the possibility of ever knowing that one thing has caused another. For the idea of one thing causing another is the idea that there is some sort of necessary connection between the two events; but we never see the necessary connection, we just see the first event and then the second.
On the other hand, as Kant saw it, Leibniz and the rationalists generally held that all knowledge comes ultimately from reason. For Leibniz, as for Hume, there was a continuum between sensations and ideas ("ideas of sense" and "ideas of reason"), but for Leibniz what was fundamental was the role of reason rather than the role of sense. The ideas of sense were for Leibniz just confused ideas of reason. This is not quite fair, but roughly the way to find out about the way things were was not to look but to sit back and think for a while. On this view we could have all sorts of knowledge about the world just by thinking about things clearly enough and not getting too confused by our sense impressions, and the knowledge we arrived at would be a priori or independent of experience. (For example, we could arrive at knowledge about God in this way.) This led to what Kant called "dogmatism," the view that reason by itself can tell us things about the world and also about things we can never have sense impressions of, like God.
Kant thought it vital to recognize, in contrast to both rationalism and empiricism, that there were two independent factors which contributed to knowledge and each of which played an essential role: a receptive faculty of sensibility, which is our faculty of receiving sense impressions, or what Kant called "intuitions," and an active faculty of understanding, which forms concepts and makes judgments about things. (Kant's remark that "intuitions without concepts are blind, concepts without intuitions are empty" is often quoted.)

Kant tried to employ this distinction in showing, against Hume, that we could have a priori knowledge about the empirical world and in particular about causality; and against Leibniz that we could not have a priori knowledge, or any other kind, about anything except possible objects of experience. For example, we could not have knowledge about God, or about the immortality of the soul, or, most importantly for our purposes, about whether or not we are genuinely free. (But this denial is not quite as extreme as it sounds: although Kant denies that we can know anything about God, freedom, or immortality, he also holds that for practical purposes we must presuppose, or "postulate," the existence of all three.) These two claims are both crucial to Kant's epistemology; the first is a kind of response to skepticism, while the second seems almost to be itself a kind of skepticism.

1. Kant's positive claim: we can have synthetic a priori knowledge

Let us examine first, then, Kant's claim that we can know things about the empirical world without basing this knowledge on experience. How can we know anything about the empirical world in this way? Very roughly, Kant's answer is: "Because we made the empirical world according to certain rules; we can know a priori those features of the world that we put there in the first place." Kant called this view his "Copernican revolution" in philosophy. This may seem odd, since in one respect Kant's view is counter-Copernican: Copernicus displaced man from his supposed place in the center of the universe, while Kant makes man in an epistemological sense more central to the universe than he had been thought to be. But the analogy Kant has in mind is with Copernicus' explanation of the observed motions of the stars in terms not of their actually moving but in terms of our moving. Similarly, Kant thinks that many of the most general features of the world we observe are present not because they are in the things themselves, but because we put them there. The explanation of space, time, causality, and other pervasive features of the universe is not to be found in the world itself, but in our own mental makeup.

This is not the place to examine in detail Kant's arguments for various sorts of a priori knowledge. But it may be helpful to look briefly at two purported examples of synthetic a priori knowledge. First, Kant thinks that our knowledge of geometry is synthetic a priori. Kant thinks of geometry as a collection of truths about space, and hence about the empirical world. (Many would disagree with him, but let's pretend he is right.) Kant thinks we could not get the kind of knowledge we have of geometry by making inferences from our experience. (Why not? Because we are absolutely certain of mathematical truths, while the most we could glean from experience would be the view that they were highly probable.) How then can we discover these truths about space? Because, Kant says, space is not a feature of the world as it is in itself, but something we impose on the world in ordering our experiences. We are simply constructed in such a way that information about the world has to come organized spatially (and temporally, but that's another story).

Without examining Kant's arguments in detail, it may be helpful to consider a metaphor he uses. Space and time, he argues, are forms of intuition. There is a distinction between the "matter" or content of our experience, on the one hand, and its form or organization, on the other, and space is part of the form rather than of the content. Compare the notion of a poetic form. Poems may come in many different forms: sonnets, sestinas, villanelles, and so on. Within a given form, we know that all poems will share certain features: all sonnets, for example, will have 14 lines, will be written in iambic pentameter, and will have a certain rhyme scheme (one of several, actually, depending on the kind of sonnet). These features are all part of the sonnet form, guidelines one must conform to in order to be writing a sonnet at all. Does the form of the sonnet, then, consist simply in all those features that all sonnets share? Not quite. Suppose all sonnets were written in English: that would not make being written in English a part of the sonnet form, since one could write a sonnet in German. The form of a sonnet consists of those features which any sonnet must share, or, as we might say: of those features shared by any possible sonnet.

Now, Kant thinks that intuitions too may come in various forms; in particular he thinks some beings may have intellectual rather than sensible intuition (know things by creating them rather than by being affected by them). But he thinks that all human intuitions share, and must share, certain features, namely that they are of spatiotemporal objects. (Well, actually, in the case of intuitions of mental phenomena, just temporal ones. "Outer sense," from which we get intuitions of external objects, is organized spatiotemporally; "inner sense," which gives us intuitions of "inner" or mental phenomena, is just organized temporally.) Since all human intuitions must be spatial and temporal, space and time constitute the form of human intuition.

Just as we know independently of the matter of a sonnet that it will have certain features, so we know a priori or independently of the matter of experience that all intuitions are of spatiotemporal things; this is why we can have a priori knowledge of geometry.

This analogy points up a further interesting point. Although we can give a general characterization of a poem's form without saying anything about its content, and a general characterization of its content without saying anything about its form, we cannot identify specific parts of a poem as formal features or matters of content. Any particular word or phrase will contribute to both the poem's form and its content; the division of the poem into form and content is not a division into parts. The same thing is true, for Kant, of experience. We cannot point to some parts of our experience as formal and some parts as material; all are matter organized by form.
But this very analogy suggests a possible disanalogy. In the case of a poem, though form and content are closely connected, we can at least roughly distinguish the formal features of the poem from its content: the fact that it is in iambic pentameter is a formal feature; the fact that it paints a pessimistic picture of love is a matter of its content. We can specify the formal features of a poem without saying anything about its content, and we can give a paraphrase of its content without saying anything about its form. (I have some misgivings about how far this latter is possible; thus my weak claim that there may be a disanalogy.) In Kant's case, however, although we can say something about the form of human intuition in isolation from its content, we seem to be able to say nothing about its content independent of its form. For example, while an object's shape and size are obviously spatial properties (and hence part of its form?), it would be a mistake to identify its color, e.g., as a part of its content, i.e. as a feature of the object itself rather than something imposed by the form of our experience. Why should this be? A partial answer might be the following. The only way we can talk about the content of a poem independent of its form is to describe its content in a different form, e.g. by giving a prose paraphrase. But precisely this is impossible in the case of experience, since Kant thinks the only form of human experience is spatiotemporal: there is no question of discussing the content of our experience in a different form, hence no way to get a handle on the matter of experience as what is shared by various experiences of different forms.

I promised two examples of purported synthetic a priori knowledge, and so far have delivered only one. The second one I want to mention is our knowledge of the law of causality, that is, of the proposition that every event has a cause. Kant thinks we can know this for something like the same reason that he thinks we can know the truths of geometry. Causality is one of twelve "Categories" Kant argues we impose on our experience. But the categories are imposed by the understanding rather than by intuition; we might think of them as "forms of thought" (to my knowledge Kant does not use quite this expression, though he does regard them as derived from the "forms of judgment"). The idea is that, just as we cannot perceive objects without perceiving them as spatiotemporal, we cannot think about them except as conforming to the Categories. Thought presupposes the categories in something like the same way perception presupposes space and time. But experience requires both concepts and intuitions, both sensibility and understanding; so experience requires both space and time and the categories.
A very quick version of Kant's argument that we must be able to apply the Categories, including the notion of causality, to our experiences might be the following. (I hope it is not seriously misleading, though it omits mention of many niceties. Kant's argument is found in the "Transcendental Deduction of the Categories" in the Critique of Pure Reason; a simpler version is in his Prolegomena to Any Future Metaphysics, sections 14-23.) Kant begins by observing that our experience is coherent and unified. But, he argues with considerable plausibility, this is only possible if we take our experiences, our mental representations, to be representations of objects external to us. Our experiences would be a mere meaningless jumble, a "blooming buzzing confusion" to use James's expression, if we were unable to take them to be experiences of an external world. But, Kant argues, we can only take our experiences to be representative of an external world if we have a general concept of an object, the notion of an "object in general." We need, that is, to have some idea what it would be for our representations to be related to something outside us. It is hard to imagine that we could acquire the relevant concept by merely attending to the play of our consciousness. Or rather, it is hard to see how we could extract this idea from our experience if we had not already placed it there, had not already used it in interpreting our experience (thereby changing the subjective quality of that experience). Nothing about the phenomenological quality of our experience makes it necessary that it be about external things; it could be just an endless meaningless play of colored shapes and noises and sensations of hot and cold, solidity and resilience. To interpret our experience as experience of an external world, we need to supply the notion of an external world, and especially of an external object. Kant suggests that the twelve Categories constitute the concept of an object in general, and thus are what we add to our experience (so to speak) in order to transform it into experience of an external, objective world. Of these twelve Categories, the two most important are causality and substance; thus, the idea of an object in general is the notion of a substance that stands in causal relations with other substances. (We remember that Locke admitted that we could not acquire an adequate conception of substance from experience, and that Berkeley insisted we could acquire no such notion at all from experience. Kant agrees with them about this much, but does not accept their assumption that experience is the only possible source of ideas. Kant agrees with the rationalists, as against the empiricists, in supposing that we possess some ideas which we do not derive from experience but which are simply supplied by our reason.)

It will be noticed that if this argument establishes something about causality, it is only that we need to be able to apply the idea to our experiences, not the general thesis that every event must have a cause. Kant does have an argument for the more specific claim, to be found in the infamous "Second Analogy of Experience," but I will not discuss it here.

This concludes my discussion of the positive side of Kant's epistemology, of his argument that we can have a priori knowledge of synthetic truths. We have seen that Kant thinks we can have such knowledge because the empirical world is partly our construction, and we construct it so as to be spatiotemporal and to consist of substances standing in causal relations. By looking inward rather than outward, we can discover those features we impose on the world. It is time to turn to the negative part of Kant's view, his rejection of any sort of knowledge of things in themselves.

2. Kant's negative claim: we can know nothing about things in themselves.

Kant's positive epistemological doctrine is explicitly designed to be a response to skepticism. Yet at the same time Kant's view includes some very skeptical-sounding claims. Indeed, although Kant argues against both Humean skepticism and Berkeleian idealism, his view often appears to me to be a rather odd combination of both.

Kant insists that all our knowledge is knowledge of what he calls appearances. Appearance is usually contrasted with reality, and so this sounds rather like a denial that we know anything about reality. It is also reminiscent of Locke's claim that all our knowledge is of the relations of ideas. But these apparent connections are at least somewhat misleading, since Kant regards the entire empirical world as a collection of appearances. Kant does not for a moment doubt that we know a great many things about objects external to ourselves; both mathematics and natural science, he believes, give us such knowledge. The conventional distinction between appearance and reality, between the way things look to us (sometimes mistakenly) and the way they really are, is for Kant a distinction within the realm of appearances.

And yet there is something ultimately subjective about appearances, and thus about the whole empirical world, for Kant. The world of appearances is partly our own creation, yet Kant thinks that standing behind this world of appearances is a really real world that we have nothing to do with, which is utterly independent of our knowledge. Our world, the empirical world, the world of appearances, is objective in the sense that it is the same for all of us: it is intersubjective. But it is subjective in that it is a uniquely human world. Kant thinks that there could be creatures with other sorts of intuition than ours. (Does he think there could be creatures with other sorts of understanding? I do not know.) Such creatures would not share our world. But something, the external source of our experience, would be the same for us and them. This, one is tempted to say, is the really real world. At any rate, this external source comprises what Kant calls the "things in themselves." A thing in itself is a thing as it is independent of any human conceptualization. And Kant argues that we can know nothing at all about things in themselves.
This suggests the following characterization of Kant's position. Think of the problem of skepticism as the problem of how, given knowledge of the way things appear to us to be, we can acquire knowledge of how they really are (and remember Descartes' linking of the appearance-reality and mental-physical distinctions, so that appearances are thought of as mental and reality as physical). Then two main responses to the problem are Berkeley's reductive response of insisting that reality is really just a subdivision of the appearances so that a properly sophisticated knowledge of the appearances is automatically also knowledge of reality, and Hume's skeptical response of accepting that reality is separate from appearance and denying that we can know anything about reality. (Well, probably this isn't quite the right characterization of Hume. Never mind.) Now, Kant seems to combine both responses! He insists with Berkeley that (empirical) reality is just a matter of appearance, so that knowledge of empirical reality is as straightforward as knowledge of appearances. This is the positive part of his doctrine. But he insists with the skeptic that we can know nothing of the really real world, the world of things in themselves. And this negative part of his doctrine, while admittedly it does not threaten any of our practices (as Humean skepticism also did not), seems as skeptical as the hardest-core skeptic could wish.

It is useful to distinguish three related doctrines of Kant's. There is, first, the doctrine that we cannot have knowledge of anything we cannot experience. Anything beyond the reach of experience is unknowable. (But we must be careful to construe the reach of experience broadly enough; we may not be able to directly experience dinosaurs or quarks, but they are systematically related to things we do directly experience, and that is good enough.) In this Kant is the heir of the empiricists and the precursor of the positivists. The second doctrine is that we can have no experience of things in themselves. Of course it follows from the first and second doctrines together that we can know nothing about things in themselves, but many who accept the first doctrine will want to reject the second. Finally, Kant's third doctrine is that God, freedom and immortality all belong to the realm of things in themselves, and thus to the realm of things about which we can know nothing. About this third doctrine I would like to assert dogmatically that I think it entirely unjustified. It seems to me that our knowledge of these matters is no less direct than our knowledge of the more theoretical parts of physics or the more remote parts of history. In my view, if there really is a distinction between the phenomenal and the noumenal, between appearances and things in themselves, then God and related matters are thoroughly phenomenal, a part of the world of appearances. --Though of course if they exist then like everything else they have noumenal underpinnings. But given that much of Kant's motivation for insisting on the unknowability of things in themselves has to do with protecting God from the failure of arguments for his existence, and with protecting freedom from the law of causality, accepting my assertion might leave him with little reason for insisting on his sharp distinction between knowable phenomena and unknowable noumena.

We have been discussing Kant's epistemology for some time. We have seen that Kant's epistemological motivation for idealism is to thwart skepticism, and we have seen that while his philosophy is in one sense antiskeptical, in another equally valid sense it seems to be very skeptical indeed. It is time now to turn to Kant's practical motivation for idealism.

C. Kant's Practical Philosophy

We have seen already that Kant's main practical motivation for idealism is the desire to protect human freedom from the threat of the very law of causality he wants to establish as an a priori law of nature. He also wants to protect the existence of God and the immortality of the soul from the lack of either evidence or successful a priori argumentation in their favor. And he thinks that protecting these things is essential to the preservation of a meaningful morality.

I would like to stress two points. First, I will discuss the way Kant tries to preserve some theoretical room for the possibility of freedom; second, I will discuss his view that we must "postulate" the existence of freedom.

How does Kant's idealism help preserve the possibility of freedom? Well, very roughly, the idea is that while the law of causality governs the empirical world, this is only because we put it there; since it is due only to our own activity, it does not govern us. Of course this representation of Kant is too crude. To be a bit more accurate, we need to note that for Kant everything, including ourselves, has a noumenal ground or aspect. We have both an empirical, phenomenal character and a noumenal character. Insofar as we are empirical, we are, like everything else, subject to the law of causality. But causality, being our own contribution to the world of appearances, does not apply to things in themselves, including our own noumenal selves.

So far, then, the picture is something like this. Any action has empirical causes. These causes themselves had sufficient causes, and so on back into the dim recesses of history. But on the other hand, each action has a noumenal ground, and causality does not apply to noumena, so each action is produced (not "caused," exactly, since that notion doesn't apply here) by something which is not itself caused.

This helps explain how Kant thinks actions can be produced by a "first beginning" rather than a "subordinate beginning" and thus how they can be free. But it cannot be the whole story. For all that has been said so far, it could be that everything is determined empirically and yet free insofar as it is noumenally produced. But though rocks, for example, have a noumenal ground, they are not thereby free. Kant's conception of freedom is richer than this.

A second element in Kant's conception of freedom is the idea that an action is free only if it is produced by reason rather than by desires or impulses. There is involved here a typically philosophical identification of our true self with our capacity as reasoning creatures, the view that thinking is essential to human nature in a way feeling is not. Thus true self-control must involve acting as reason dictates, not as our urges incline us. Our desires and inclinations are in a sense extraneous to us, so that to let them control our actions is in a way to give up our own control over them and to let them be determined by caprice. This seems to be the picture involved, for example, when Kant speaks of freedom as "independence of coercion through sensuous impulses" (Kemp Smith, p. 465; my emphasis).

This picture of a free life as one governed entirely by reason rather than impulse seems quite intelligible, and provides a kind of rationalist counterpoint to the Humean view that "reason is, and ought to be, the slave of the passions." But Kant links this view with the phenomena-noumena distinction in a rather odd way. He takes the view that reason is a noumenal faculty, by contrast to inclination, which is phenomenal. Thus to have one's actions determined by desires is to have them empirically caused, while to have them determined by reason is to have them noumenally produced. Thus to act as reason dictates is to open up the possibility of freedom, while to act as one's impulses lead is necessarily to be unfree. Kant sometimes seems to picture reason and desire battling for control of the will, with one's freedom or lack of freedom hinging on the outcome.

This all fits very oddly with other parts of Kant's view. Surely it is not as though there can be a sort of competition for the will: every empirical phenomenon has an empirical cause, so in particular every action, viewed as an empirical phenomenon, must have a sufficient empirical cause. Doing what reason dictates, having one's action noumenally produced, can hardly make it the case that it is not phenomenally produced. And it also seems odd to regard reason as somehow noumenal. There is, to be sure, a plausible connection between reason and noumena: reason, for Kant, is the faculty that tries unsuccessfully to acquire knowledge of things in themselves, and that more successfully guides our action by producing ideals which are also in a sense noumenal, since they are pictures of the way things should be, and this is something we do not and could not experience or infer from experience. Thus we can say that the subject matter of reason is noumenal, while the subject matter of sensibility and understanding is phenomenal. But surely it does not follow that the faculties themselves are divided in this way; nor does it follow, therefore, that acting for reasons is having one's action noumenally produced in contrast to having it phenomenally produced. 

(But perhaps I am being unnecessarily uncharitable here. Perhaps Kant does not mean the notion of noumenal production to conjure up images of a spooky "causation" from beyond the pale. Perhaps he can regard acting on reasons as a kind of noumenal origination whose vehicle is an empirical psychological faculty. Then acting on reasons and acting out of inclination would both be cases of having one's action empirically caused, but the former would also be a case of one's action being influenced by noumena, namely ideals, in a way the latter would not. Of course, in both cases action would have a noumenal ground; our empirical actions and their causes would still somehow be appearances of an underlying realm of things in themselves. Nevertheless noumena would be relevant to actions done from reason in a way they would not be to actions done out of inclination, since in one case but not the other the empirical causes of the action would count as a consideration of noumenal matters. (But notice that counting both ideals and things in themselves as noumenal seems to stretch the term "noumenal" uncomfortably broadly.) This interpretation would, at any rate, remove the apparent inconsistency between Kant's view that every action is empirically caused and his view that one is free only if one acts out of reason rather than inclination. But it seems to do so only at the expense of the view that the cause of a free action is a "first beginning.")

What is right in Kant's curious compatibilism? Here is one suggestion. What is right is a certain distinction between two points of view we can take up with regard to a subject matter, or two vocabularies we can use to describe things. Everything has two aspects, for Kant, a phenomenal one and a noumenal one; causality pertains to the phenomenal aspect, freedom to the noumenal one. We might describe the vocabularies appropriate to these two points of view--that of causes in the one case, and that of reasons in the other--as incommensurable. And something about this view is very tempting.

A number of contemporary philosophers (especially Donald Davidson and Jerry Fodor) would maintain something rather similar, namely that the language of action ascription and the language of physics do not mesh. Assume that we are physical and that determinism is true: it does not follow that there is a physical explanation for why we act as we do, even though there may well be a physical explanation for why our bodies move the way they do. Causal explanation seems to be closely tied to causal laws: to give a causal physical explanation of an action, it seems, we would have to have a causal law with physical facts as antecedents and facts about actions as consequents. But this seems unlikely, because although every action is a physical event, there seem to be no law like connections between kinds of physical event and kinds of action. (Try to give a physicalistic analysis of voting, or paying a bill, or even smiling.) The kind of language we use in describing the reasons someone did something, and the kind of language we use when constructing physical causal explanations of phenomena, both apply to the very same things--but they fail to mesh, there is no translation from expressions in the one vocabulary to expressions in the other. Freedom is a notion that makes sense relative to the vocabulary of reasons, but not relative to the vocabulary of physics, so there is at least the possibility of a kind of compatibilism according to which people are both free (considered as rational agents) and determined (considered as collections of molecules). This kind of view is reminiscent of Kant's, but with two differences: first, on this view we can know about both sorts of classification; reasons are not relegated to some unknowable noumenal realm; second, in a sense the physical classification may be more rather than less basic: in the fairly weak sense, namely, in which the mental supervenes on the physical, that is, in which there can be no mental difference without a physical difference.

Let us return from this brief digression. We have been discussing Kant's attempt to make theoretical room for freedom in a deterministic universe, and especially the role his idealism plays in this attempt. But we need to remember that Kant does not think that we can know that his version of freedom ever obtains. Freedom is supposed to consist in the noumenal production of action, and we can never know anything about the noumena, so we can never know when if ever freedom obtains. All Kant thinks he can do is to show the theoretical possibility of freedom.
Does this mean that we should regard it as an open question whether we are free or not? Well, theoretically it is an open question. But Kant thinks that if we are to act at all we need to suppose that we are free. There are no theoretical arguments to show that we are, but in order to act, and especially in order to act morally, we need to presuppose or "postulate" that we are genuinely free. (Kant also thinks we need to postulate that God exists and that our souls are immortal, but his arguments for these further postulates are quite unconvincing -- e.g. that we need to aim at perfection, but cannot aim at anything we regard as impossible, and hence, since perfection obviously is not attainable in the span of a human lifetime, we need to assume that we have much longer than that to work on it.)

Since Kant thinks that we need, for practical purposes, to believe things that we can have no theoretical evidence for, he believes in what he calls the "primacy of practice": in cases where practice requires a belief that theory will not justify, we need to go ahead and adopt the belief. Thus the requirements of practice take precedence over the requirements of theoretical reason. We might think of this theme in Kant as a kind of anticipation of pragmatism; certainly it provides the basis for Fichte's more sweeping claims for the primacy of practice, and some of Fichte's views have a very pragmatic flavor.

I have presented Kant's transcendental idealism as a kind of uneasy compromise between a reductive Berkeleian idealism on the one hand and a wholesale skepticism on the other. This may be unfair to Kant, though I do not personally think it seriously unfair; but at any rate it is useful for dialectical purposes, since later German idealists, and especially Hegel, saw things in much this light. What gives Kant's view its skeptical quality is his view that behind the appearances are things in themselves about which we cannot know, and it was precisely this feature of his view that the later idealists found most indefensible.

 Last update: April 6, 2009. 
Curtis Brown  |  Classical Modern Philosophy  |  Philosophy Department  |  Trinity University