B. Arguments or Reasons?

Clearly, there are problems of coherence here. Suppose we ignore them for the moment: what kinds of reasons does Kant give for the contention that we can’t think about, refer to, predicate properties of the Dinge? Or, if he gives no such reasons (perhaps because he thinks we can think about them), what sorts of reasons or arguments does his work suggest for that conclusion? This conclusion—that our concepts are really rules for synthesizing the manifold into phenomenal objects and that the only things we can think about are objects we ourselves have somehow constructed—is, to say the least, rather startling. Some pretty powerful arguments would be required.

Argument for this view is distressingly scarce. It is extremely difficult to find much that could pass muster as an argument, or even as one of those “considerations determining the intellect” John Stuart Mill sometimes gave when, as he conceded, he didn’t have an argument. There is nothing here like the ontological or cosmological arguments for the existence of God, or Descartes’ argument that a person is not identical with her body (but is, instead, an immaterial substance), or the argument for the conclusion that propositions, the things we believe and assert, are not contingent objects.²⁷²⁷   See my Warrant and Proper Function (WPF), pp. 117ff. Perhaps one must think of the radical subpicture as a sort of hypothesis proposed as best explaining certain phenomena. More likely, those who urge it are simply overwhelmed by what they see as its sheer intellectual beauty and power; they don’t feel the need of argument. Indeed, they find the picture so dazzling they are willing to put up with a strong dose of incoherence in addition to absence of argument. Well, if you find the radical subpicture overwhelmingly attractive, then (incoherence aside) I guess you’ll have to go with it. Then again, that doesn’t constitute much of a reason for the rest of us—those of us more impressed by the incoherence of the picture than its beauty—to accept it.

There is, however, a set of Kantian considerations that some might see as taking us partway to the conclusion. These are to be found in what he says about the antinomies: allegedly powerful arguments on both sides of a given question. Thus there is an allegedly compelling antinomical argument for the thesis that the world had a beginning in time, but an equally compelling argument for the antithesis that it did not. In the same way, there are compelling arguments for the theses that the world is composed of simples, that there is such a thing as agent causation (where an agent cause is a being that freely originates a new causal series), and that there is an absolutely 22necessary being; sadly enough, however, there are equally compelling arguments for the antitheses that the world is not composed of simples, that there is no such thing as agent causation, and that there is no absolutely necessary being. Here we seem to be in a nasty fix; we can prove four (everything in the Critique comes in fours) important theses, and for each of these four, we can also prove its denial.

Now Kant apparently intends these antinomies to constitute an essential part of the argument for his transcendental idealism, the doctrine that the things we deal with (stars and planets, trees, animals and other people) are transcendentally ideal (depend upon us for their reality and structure), even if empirically real. We fall into the problem posed by the antinomies, says Kant, only because we take ourselves to be thinking about things in themselves as opposed to the things for us, noumena as opposed to mere appearances:

If in employing the principles of understanding we do not merely apply our reason to objects of experience, but venture to extend these principles beyond the limits of experience, there arise pseudo-rational doctrines which can neither hope for confirmation in experience nor fear refutation by it. Each of them is not only in itself free from contradiction, but finds conditions of its necessity in the very nature of reason—only that, unfortunately, the assertion of the opposite has, on its side, grounds that are just as valid and necessary. (A421, B449)

We solve the problem by recognizing our limitations, realizing that we can’t think, or can’t think to any good purpose, about the Dinge.

In presenting the antinomies, Kant does not explicitly argue for the radical subpicture. But suppose we try to find something like an argument there, either for the radical subpicture or for the conclusion we have been deriving from the radical subpicture, the conclusion that our concepts do not apply to the noumena, so that we cannot refer to and think about them. Perhaps the premises would be:

(2) If we are able to think about and refer to the Dinge, then the premises of the antinomical arguments (the premises of the arguments for the theses and for the antitheses) are about the Dinge and are all true,

and

(3) If those premises are all true, then the theses and antitheses would all be true, so that contradictions would be true.

Naturally enough, however,

(4) No contradictions are true.

Therefore:

(5) We cannot think about or refer to the Dinge.

We could perhaps weaken the first premise (2) to make it a bit more plausible:

(2*) If we can refer to and think about the Dinge, then each of the premises of the antinomical arguments will be about the Dinge and have overwhelming intuitive support.

(This is weaker, of course, because it says, not that the antinomical premises are true, if we can think about the Dinge, but that they strongly seem true to us.)

The second premise would then be:

(3*) If each of the premises has overwhelming intuitive support, we will have overwhelming reason to accept each of the theses and antitheses, and we see that each thesis is contradicted by its antithesis.

If, however, we weaken the first premise, we must strengthen one of the other two. Perhaps we could strengthen the third as follows:

(4*) It couldn’t be that we should have overwhelming reason to accept a proposition p and also its contradictory not-p.

And the conclusion would be as before.

Is it really true that (as (4*) claims) we couldn’t have overwhelming reason to accept both a proposition p and also its denial not-p?²⁸²⁸   It seems we could have good reason to accept each member of a set S of beliefs such that there is no possible world in which all the members of S are true (the conjunction of the members of S is impossible), as is shown by the paradox of the preface. I write a book, of course believing every proposition asserted therein. Past experience and self-knowledge, however, lead me to think that very likely the book contains at least one false statement. (All of my previous books, as I’ve discovered to my sorrow, contain false statements.) In the preface, therefore, I sadly concede that at least one statement in the book is false. The total set of my beliefs, therefore—the statements in the book plus the statement that at least one statement in the book is false—is such that it must contain at least one falsehood; nevertheless, I have good reason to accept each member. This would be an interesting inquiry but would take us too far afield; in any event, it isn’t necessary for our present purposes, for there are at least two impressive problems with these arguments, one debilitating and the other fatal. I shall briefly outline the first and then look into the second in more detail. The first, the debilitating objection, is that even if we are not able to think of the noumena, we can think of the phenomena; and if the first premises of these arguments are true for the noumena, what is to prevent their being true for the phenomena 24as well? The two versions of the first premise ((2) and (2*)) of the argument claim the following: if it is true that we can think about the noumena, then the antinomical premises are about the noumena and either are true or have overwhelming intuitive support. Isn’t it equally apparent that if we can think about the phenomena, then the antinomical premises are about the phenomena and are either true or have overwhelming support? If so, however, the argument would also prove that we can’t refer to the appearances. What it would really prove, then, if it proved anything, is that we can’t refer to or think about either noumena or phenomena. Because noumena and phenomena are all the things there are, the conclusion would be that we can’t think about anything; and that seems a bit strong.

Much more should be said about this objection to the argument, but I want to turn to the fatal objection. That is just that the antinomical arguments are not, to put the best face on it, at all compelling. Here I will argue this only for the premises of the first antinomy; exactly similar comments would apply to the others. In the first antinomy, there is an argument for the conclusion that “The world had a beginning in time and is also limited as regards space” (A426, B454); this is the thesis. There is also an argument for the antithesis: “The world has no beginning, and no limits in space; it is infinite as regards both time and space” (A426, B454). And the idea (in accordance with premises (2) and (2*)) is that if we can think about and refer to the Dinge, then both of these would be true or would have overwhelming intuitive support.

Well, what is the argument? I am sorry to say it is hard to take seriously. The argument for the thesis goes as follows:

If we assume that the world had no beginning in time, then up to every given moment an eternity has elapsed, and there has passed away in the world an infinite series of successive states of things. Now the infinity of a series consists in the fact that it can never be completed through successive synthesis. It thus follows that it is impossible for an infinite world-series to have passed away, and that a beginning of the world is therefore a necessary condition of the world’s existence. (A426, B454)

This argument proceeds by reductio ad absurdum: show that the denial of your conclusion leads to a contradiction, thereby proving your conclusion. The first premise is that if the world had no beginning in time, then at any point in time an infinite stretch of time would already have elapsed. This is dubious because it is at least abstractly possible that time and the world began together, some finitely many years (or seconds) ago. If so, then we should say that the world didn’t have a beginning in time, although it did have a beginning with time. But let that pass. According to the second premise, “the infinity of a series consists in the fact that it can never be completed through successive synthesis”; that is, it is characteristic of an infinite series that 25it can’t be completed by starting from the beginning (or, more generally, some point only finitely far from the beginning) and adding things (events, say) one at a time (or more generally, finitely many at a time). This is true, provided the things (events) in question are added at a constant rate. If you start with the first event (or the nth, for some finite n) and add another event every second, you will never complete the series: at any subsequent time only a finite number of events will have occurred. According to current lore about the infinite, however, there is no bar of this kind to completing the infinite series in a finite time if the time taken for each event diminishes appropriately. For example, the first event takes one second to happen; the second event takes half a second; the third a quarter, the fourth an eighth of a second, and so on. At that rate, it won’t take long at all for an infinite number of events to have elapsed—only a couple of seconds.

But the real problem with the argument lies in a different direction. Kant points out that an infinite series can’t be completed by starting from some point finitely far from the beginning and adding members finitely many at a time at a constant rate; fair enough. He then concludes, “It thus follows that it is impossible for an infinite world-series to have passed away, and that a beginning of the world is therefore a necessary condition of the world’s existence.” This doesn’t follow at all. To claim that it does is to claim just what is to be proved: that the series in question had a beginning. The premise tells us that if you start from some finite point in a series—that is, some point finitely far from the beginning of the series—and add a finite number per unit time, then you will never complete the series. Fair enough; but if the world has existed for an infinite stretch of time, then there was no first moment, no first event, and no beginning either to the series of moments or the series of events; more generally, at any preceding moment an infinite time would already have elapsed. To conclude, as Kant does, that it is impossible that an infinite series of events has occurred is just to assume that the series in question had a beginning—that is, is finite—but that is precisely what was to be proved. So the argument really has no force at all. It is not as if it is an argument the premises of which have a certain limited amount of intuitive plausibility; it is rather that this transition to the conclusion completely begs the question by assuming what was to be proved: that the series in question has a beginning. The argument therefore fails to establish its conclusion; it merely assumes it. It therefore gives us no reason at all for accepting that conclusion.

The argument for the antithesis is no more promising. Here is how Kant puts it:

Let us assume that it [the world] had a beginning. Since the beginning is an existence which is preceded by a time in which the thing is not, there must have been a preceding time in which the world 26was not, i.e., an empty time. Now no coming to be of a thing is possible in an empty time, because no part of such a time possesses, as compared with any other, a distinguishing condition of existence rather than nonexistence. . . . (A427, B455)

Again, the argument is by reductio: assume the denial of your conclusion and show that it is impossible, thereby establishing the conclusion. Here the two premises are

(6) The beginning of an event or a thing is always preceded by a time in which the thing is not, that is, a time at which the thing in question does not exist.

and

(7) In an empty time (a time at which nothing exists) nothing could come to be, because there would be no more reason for it to come to be at one part of that empty time than at any other part of it.

Neither premise is at all compelling. As to the first, this is true only if it is not possible that time and the world (the first event) should come into existence together, simultaneously. Is it known that this isn’t possible? Certainly not. Indeed, some of the most popular theories of time (relational theories) would assume, not merely that this is possible, but that it is true.

As for the second premise, it is equally unpromising. Suppose (in accord with the picture governing the argument) an infinity of time had elapsed before the first event of the world took place—before its creation, say. The objection is that there would have been no more reason for God to create the world at one moment than at any other; hence he wouldn’t or couldn’t have created it at any moment at all. Again, why believe this? If God proposed to create the world, and no time was more propitious than any other, why couldn’t he just arbitrarily select a time?²⁹²⁹   Compare Augustine’s answer to those who wanted to know what God was doing before he created the world in The Confessions of St. Augustine, tr. Rex Warner (New York: New American Library, 1963), book 11, chapter 12, pp. 265–66.

This argument is like those arguments that start from the premise that God, if he created the world, would have created the best world he could have; they go on to add that for every world God could have created (weakly actualized,³⁰³⁰   For the notion of weak actualization, see my The Nature of Necessity (Oxford: Clarendon Press, 1974), p. 173, and Alvin Plantinga, ed. James Tomberlin and Peter van Inwagen (Dordrecht: D. Reidel, 1985), p. 49. say) there is an even better world he could have created or weakly actualized; therefore, they conclude, he wouldn’t 27have weakly actualized any world at all, and the actual world has not been weakly actualized by God. Again, there seems no reason to believe the first premise. If there were only finitely many worlds among which God was obliged to choose, then perhaps he would have been obliged, somehow, to choose the best (although even this is at best dubious).³¹³¹   See Robert Adams’s “Must God Create the Best?” Philosophical Review 81 (1972), pp. 317–32. But if there is no best world at all among those he could have chosen (if for every world he could have chosen, there is a better world he could have chosen), why think a world’s failing to be the best is sufficient for God’s being unable to actualize it? Suppose a man had the benefit of immortality and had a bottle of wine that would improve every day, no matter how long he waits to drink it. Would he be rationally obliged never to drink it, on the grounds that for any time he might be tempted to, it would be better yet the next day? Suppose a donkey were stranded exactly midway between two bales of hay: would it be rationally obliged to stay there and starve to death because there is no more reason to move to the one bale than to the other?

The arguments for the other antinomies don’t fare any better. In no case is there anything like a conclusive argument (given the assumption that we are thinking about the Dinge) for either the thesis or the antithesis. In some cases, we may not know or be able to tell which (thesis or antithesis) is true: but that doesn’t constitute much of an argument for the conclusion that we can’t think about the noumena. What would be needed for the argument to work would be a really powerful argument for the thesis and an equally powerful argument for the antithesis. In none of these cases do we have something like that.

Suppose we think a bit further about antinomies and paradoxes in connection with this question of concluding that we simply can’t think about a given area or topic. Consider the Russell paradoxes, in their simple set-theoretical guise. Like Frege, we are all initially inclined to think that for every condition or property, there exists the set of just those things that meet the condition or have the property. It is pointed out that there is such a property as being nonselfmembered, the property a thing has just if it is not a member of itself; hence there must be a set S of nonselfmembered sets, but then S is a member of itself if and only if it is not a member of itself, which is a contradiction. Here it would be unduly enthusiastic to conclude that we can’t really think and talk about sets as they are in themselves and can instead think only about sets that we have ourselves constructed, sets as they appear to us. One takes the argument as proving only that there is no set of nonselfmembered sets and that, contrary to appearances, it is not true that for every property or condition, there exists the set of just those things satisfying the condition or displaying the property.

Take, instead, the Russell paradox as specified to properties, rather than sets; in some ways, this is a more serious paradox. One is initially inclined to think that there are properties, that some properties (for example, the property of being a property) exemplify themselves, so that there is such a property as self-exemplification, and that every property has a complement. These together lead to trouble: they imply that there is such a property as non-self-exemplification, which inconsiderately both does and doesn’t exemplify itself.³²³²   If you balk at such properties as self-exemplification and non-self-exemplification, conduct the argument instead in terms of conditions; see Tomberlin and van Inwagen, Alvin Plantinga, p. 320. Once again, however, it hardly seems to follow that we simply can’t think and talk about properties an sich. We needn’t hold that if we can think about properties an sich, then there is a property that both does and doesn’t exemplify itself. We can quite properly conclude, instead, that one of the group of propositions we are initially inclined to accept must be false, and we look for the one with the least intuitive warrant or support, the one we are least strongly inclined to believe. (We might be inclined to think, for example, that there really isn’t such a property as non-self-exemplification [even though it seems as if there is] so that either there is no such property as self-exemplification, or it is false that every property has a complement.) This is mildly disquieting, and gives us reason for a bit of humility with respect to the deliverances of reason, but we certainly aren’t forced into the position of holding that we can’t refer to and think about properties an sich.

In what conditions would this drastic conclusion be right? Perhaps in none at all, and if in some, it is hard to say which. At the least, however, it would involve our being very strongly inclined to accept each member of a set of propositions about some subject matter, which set (by argument forms we are very strongly inclined to accept) entails a contradiction. It would also involve there being several such sets of propositions about the subject matter in question. Each of the premises and arguments involved would have to have very powerful, maximal or near maximal intuitive support; otherwise, we could more reasonably hold that a premise (or argument form) with only moderate intuitive support is false (or invalid). If there were several such sets of propositions—about properties, say—and each of these propositions and argument forms had the degree of intuitive support enjoyed by, say, 2 + 1 = 3 and modus ponens, then perhaps the right conclusion to draw would be that either there simply aren’t any such things as the objects in the alleged realm, or that if there are, we are incapable of thinking about them.

Even here, however, there would be reason to doubt the success of the argument. It would involve as a premise something like:

(8) If there are several sets of premises about properties, each member of each set having maximal intuitive warrant, and the members of each set together entail a contradiction, then we cannot refer to and think about properties an sich.

The next premise would be the antecedent of (8), and the conclusion would be the consequent of (8)—that is, the proposition that we cannot refer to and think about properties an sich. But if that conclusion were true, how could we grasp (8), the first premise? That premise seems to be, among other things, about properties an sich, and if we grasp it, we are able to think about properties an sich. The argument appears to be self-referentially self-refuting: if it is a successful argument, its first premise is both about noumena and such that we can grasp it, in which case that premise must be false.

The sensible Kantian conclusion, so it seems to me, is that if, indeed, we can refer to and think about the Dinge, reason alone doesn’t tell us such things as whether the world had a beginning in time or whether there are simple substances. It seems more likely than not, perhaps, that there are simple substances and that there are free agents who initiate new causal chains in the world, but the negations of these propositions are not demonstrably mistaken. Most certainly, it is not the case that both these propositions and their denials are demonstrable, so that each is both demonstrably true and, furthermore, demonstrably false.

We must also recall that the whole scheme, the whole radical subpicture, seems incoherent in a familiar way. One who states and proposes this scheme makes several claims about the Dinge: that they are not in space and time, for example, and more poignantly, that our concepts don’t apply to them (applying only to the phenomena), so that we cannot refer to or think about them. But if we really can’t think the Dinge, then we can’t think them (and can’t whistle them either); if we can’t think about them, we can’t so much as entertain the thought that there are such things. The incoherence is patent.

Would it be possible to induce coherence by refusing to make the distinction between phenomena and noumena, speaking only of what, if we did make that distinction, would be the phenomena, and claiming that whatever there is, is either a bit of experience or an object constructed by us from bits of experience by way of concepts (i.e., rules for constructing things from experience)? That is extremely hard to believe: are the stars, for example, which, as far as we can tell, existed long before we did, either bits of human experience or objects constructed by us from bits of human experience? How are we supposed to make sense of that? On this view, furthermore, the objection to Christian belief would not be that serious Christians improperly take it that they can refer to God; the objection would be that there is no God. If there were such a person, he certainly wouldn’t be either a bit of human experience or something we have constructed from it. Still further, on this picture we ourselves (because we are among the things there are) would either have constructed ourselves from bits of experience or we would just be bits of experience; but of course we couldn’t have constructed ourselves before we existed, so we must have started off, at least, as bits of experience with 30the power to construct things. Not a pretty picture. And even if we could somehow induce coherence here, why should we feel obliged to believe it? What possible claim could such a bizarre scheme have on us?

By way of conclusion then: it doesn’t look as if there is good reason in Kant or in the neighborhood of Kant for the conclusion that our concepts do not apply to God, so that we cannot think about him. Contemporary theologians and others sometimes complain that contemporary philosophers of religion often write as if they have never read their Kant. Perhaps the reason they write that way, however, is not that they have never read their Kant but rather that they have read him and remain unconvinced. They may be unconvinced that Kant actually claimed that our concepts do not apply to God. Alternatively, they may concede that Kant did claim this, but remain unconvinced that he was right; after all, it is not just a given of the intellectual life that Kant is right. Either way, they don’t think Kant gives us reason to hold that we cannot think about God.