. 7
( 9)


neously through transcendental freedom. If there is no ¬rst cause, then
the series extends in¬nitely. The Third Antinomy, then, represents a
version of the traditional dispute over freedom and determinism.
Finally, the Fourth Antinomy concerns “The absolute complete-
ness of the dependence of the existence of the alterable in appear-
ance” (A415/B443). The question is whether something in appearance
exists necessarily. The relevant modal concept is contingency, because
“the contingent in existence always has to be seen as conditioned,”
since it refers “to a condition under which it is necessary” (A415/B442).
Because the contingent is dependent, only the absolutely necessary
could explain all contingency. So the Fourth Antinomy concerns
whether there is some absolutely necessary being in appearance.
Now the reader might wonder how the third and fourth arguments
differ from the proofs in rational theology. After all, the Third
Antinomy concerns an unmoved mover, and the Fourth Antinomy
the idea of a necessary existence, ideas employed in the cosmological
and ontological proofs. The cosmological arguments differ from the
Transcendental illusion II
theological proofs, however, because the “¬rst cause” and “necessary
being” of the Antinomies, unlike God, are located in the sensible
Kant adopts a “skeptical” resolution of the Antinomies, as opposed
to the “critical” solution of the Paralogisms. The antinomial disputes
draw apparently contradictory conclusions. The skeptical method
resolves the debates by showing that the con¬‚ict is “dialectical,”
that the conclusions are only apparent contradictories. Following the
distinction between mathematical and dynamical categories, Kant
adopts one mode of resolution for the ¬rst two “mathematical” Anti-
nomies and another for the “dynamical” Antinomies. For the former,
Kant adopts a “both false” solution, claiming that the conclusions are
actually contraries. For the latter his solution takes a modi¬ed “both
true” position, with the thesis possibly true of things in themselves,
and the antithesis necessarily true of appearances.
Given this skeptical resolution, Kant claims that the ¬rst two Anti-
nomies yield indirect support for transcendental idealism. He could
not do this with the Paralogisms, because the critical method assumes
that things in themselves or objects in general are unknowable. In the
Antinomies, Kant argues that if transcendental realism were true, then
the disjunctions at issue would have to be true. That is, the world of
appearances must be either in¬nite or ¬nite in space and time, and
the real would have to be either in¬nitely divisible or composed of
ultimate indivisible parts. From the realist standpoint the conclusions
are clearly contradictories and must have opposing truth values. But
Kant also believes the con¬‚icting conclusions are each supported by
a valid argument. Thus contradictory conclusions would apply to
the world in itself. Reasoning by modus tollens, since no object can
have contradictory properties, the world of appearance cannot be the
world in itself. Hence transcendental realism is false, and transcen-
dental idealism is true.
Some commentators reject this reasoning, interpreting the argu-
ments as depending on “veri¬cationist” claims concerning what can
be known.7 If this were correct, the arguments could not support
7 For commentators reading at least some of Kant™s arguments this way see Strawson, The
Bounds of Sense, 155“61 and 200; Posy, “Dancing to the Antinomy,” 83ff; Allison, Kant™s
Transcendental Idealism, 46“7, 312“13; and Guyer, Kant and the Claims of Knowledge, 407. I
address this issue below.
Transcendental illusion II 231
transcendental idealism, for the following reason. If the premises
concerned only what we could know about appearances, then they
could not establish what the conditions of appearance must in fact be.
The conclusions, then, would not be realist in nature, but only epis-
temological. Now inherent contradictions in reason could support
transcendental idealism only if they follow from realist claims. From
Kant™s point of view, the arguments cannot be veri¬cationist in nature.
A veri¬cationist reading appears plausible because the premises
refer to the empirical regress involved in synthesis. And we have seen
that Kant™s theory of synthesis explains cognition of spatiotemporal
objects. I agree with several commentators, however, that this is mis-
leading.8 I will argue that claims about synthesis here refer to the
intellectual procedure for thinking the world-whole, which presup-
poses Kant™s distinction between “analytic” and “synthetic” wholes.
Kant believes realists must assume that things in themselves are syn-
thetic wholes, composed of independently existing parts. If statements
about synthesis refer to manner of thinking the totality of appearances
rather than knowing them, then the arguments are not veri¬cationist
in nature.

2 . the a rgu ments of th e a nt in om ie s
The Antinomies concern what must be true of the world of appear-
ances as a whole in itself. Transcendental realists assume that this
world is “given” or exists independently of the process of knowing or
thinking it. All the arguments employ the reductio method, claiming
that the truth of the opposing view leads to a contradiction. This
is an effective way of highlighting the internal con¬‚ict of reason. As
Sebastian Gardner points out, if transcendental realism were true,
exactly one of the contradictory conclusions must be true. But since
both arguments are valid, even if we knew that a thesis were true, we
could not see “how it is possible for the antithesis to be false.”9 Here
I follow Kant™s order, discussing the arguments ¬rst and then their
8 Commentators who reject the veri¬cationist reading include Melnick, Space, Time, and
Thought, Grier, Kant™s Doctrine of Transcendental Illusion, and Gardner, Routledge Philosophy
Guidebook to Kant.
9 See Gardner, Routledge Philosophy Guidebook to Kant, 251.
Transcendental illusion II
A. The First Antinomy: the composition of the world in time and space
The First Antinomy concerns whether the world is ¬nite or in¬nite
in time and space. The thesis argues for ¬nitude: that there was a ¬rst
state of the world in in¬nite time, and that the world is bounded in
in¬nite space. The antithesis denies both conclusions, maintaining
that the world extends in¬nitely in both time and space. Kant offers
separate arguments on each side for the temporal and spatial nature
of the world. Here are the thesis arguments.
Thesis: “The world has a beginning in time, and in space it is also
enclosed in boundaries” (A426/B454). The ¬rst paragraph argues for
the ¬rst part as follows:
1. Assume the contradictory, that the world has no beginning in time.
2. By hypothesis, at any given time “an eternity has elapsed, and hence
an in¬nite series of states of things in the world . . . has passed.”
3. The idea of an in¬nite series is the idea of a succession that cannot
be completed.
4. By 3, “an in¬nitely elapsed world-series is impossible . . .”
5. By 4, the series of past states of the world in time must be ¬nite.
6. Therefore, by 1, 2, and 5, “a beginning of the world is a necessary
condition of its existence.”
The argument that the world is ¬nite in space follows in the next
paragraph at A427“8/B455“6:
1. Assume the contradictory, that the world is “an in¬nite given whole
of simultaneously existing things” in space.
2. The only way to think “the magnitude of a quantum that is not
given” as bounded in intuition is “through the completed synthesis,
or through the repeated addition of units to each other.”
3. By 2, to think the whole world ¬lling space would require com-
pleting “the successive synthesis of the parts of an in¬nite world,”
which entails that “an in¬nite time would have to be regarded as
having elapsed.”
4. But it is impossible to think of an in¬nite time as having elapsed.
5. Therefore, by 3 and 4, “an in¬nite aggregate of actual things cannot
be regarded as a given whole, hence cannot be regarded as given
6. “Consequently, a world is not in¬nite in its extension in space,
but is rather enclosed within its boundaries.”
Transcendental illusion II 233
Clearly the second argument incorporates the ¬rst by translating the
idea of a spatial whole in terms of a temporal series. Both arguments
reject an actually in¬nite world-whole because the idea of a completed
in¬nite series is impossible. Contrary to some interpreters, the alleged
impossibility is not psychological (nor epistemological) but logical.10
The main questions are why an in¬nite spatial world must be thought
through an in¬nite temporal series, and why an in¬nite completed
series is logically impossible. Answers to these questions will explain
why this argument does not apply to Kant™s theory of space and time,
as well as responding to some objections to the argument.
Kant explains his notion of the in¬nite by contrasting his “true
(transcendental) concept of in¬nity” (A432/B460) with the dog-
matist™s “defective concept of the in¬nity of a given magnitude”
(A430/B458). According to the latter, “a magnitude is in¬nite if none
greater than it . . . is possible.” That is, the defective concept of in¬n-
ity represents a maximally great magnitude. But because there is no
limit to the addition of units, there is no greatest multiplicity. The
true or mathematical notion of in¬nity, by contrast, “thinks only of
the relation to an arbitrarily assumed unit, in respect of which it is
greater than any number” (A432/B460). Unlike the defective idea
of a maximum magnitude, the true notion is of a magnitude that
surpasses any ¬nite number. Now this (true) notion is not in itself
logically incoherent. The contradiction in the notion of an in¬nite
whole arises only when it is represented as a completed series: “The
true (transcendental) concept of in¬nity is that the successive synthe-
sis of unity in the traversal of a quantum can never be completed”
(A432/B460). That is, the successive enumeration of an in¬nite series
(such as the natural numbers) can never be completed, because no
matter where one stops, there is always an additional member of the
series to be thought.
As Melnick points out “Kant is here defending the concept of an
actually in¬nite whole (= a whole encompassed by units only if these
units are together ˜greater than all number™).”11 He notes that Kant
made the same distinction between true and defective notions in the
Inaugural Dissertation of 1770. There, in a footnote in section 1,
10 Two commentators who take the impossibility as epistemological or psychological are Guyer,
Kant and the Claims of Knowledge, 407, and Kemp Smith, Commentary, 485.
11 Melnick, Space, Time, and Thought in Kant, 331. I am heavily indebted to Melnick™s inter-
pretation of the mathematical Antinomies.
Transcendental illusion II
Kant says a non-human understanding “might distinctly apprehend
a multiplicity at a single glance, without the successive application of
a measure.”12 In short, there is nothing inherently contradictory in
the idea of an actual in¬nite whole; the contradiction is in the idea
of a completed in¬nite series.
Since time is by its nature successive, the true notion of the in¬nite
entails that an in¬nite series of times cannot be thought as completed.
But it is not clear why the spatial parts of the world-whole must be
represented successively. In the last paragraph of the remark, Kant
says to think the totality of a simultaneous in¬nite extension, where
the boundaries are not given in intuition, requires having a concept
that “must establish the possibility of a whole through the successive
synthesis of the parts. Now since this synthesis has to constitute a
series that is never to be completed, one can never think a totality
prior to it and thus also through it” (A432/B460). Here Kant claims
that the idea of a whole composed of parts could arise in only two
ways: either through intuition or through thinking its relation to
its parts. Because we cannot intuit the whole of appearances, that
leaves only the second option, representing the whole by thinking its
relation to the parts. Further, Kant assumes that the thought of the
world in itself must represent the whole as composed of previously
given parts. In Kant™s terms, the world-whole in itself is a synthetic
whole, a totum syntheticum rather than an analytic whole, a totum
analyticum. As Allison puts it, “the concept of a totum syntheticum
is here operationally de¬ned in terms of the intellectual procedure
through which it is conceived . . . The problem, then, is that the
rule or procedure for thinking a totum syntheticum clashes with the
rule or procedure for thinking an in¬nite quantity.”13 Nevertheless,
Allison questions why it is necessary to conceive the series of states of
the universe as a totum syntheticum.
Melnick defends Kant by emphasizing the realist view of the mark-
ing procedures for synthetic wholes.14 Regardless of how the parts are
12 On the Form and Principles of the Sensible and Intelligible World, Theoretical Philosophy,
1755“1770, 379.
13 Other commentators refer to Kant™s distinction between a totum syntheticum and a totum
analyticum. See Kemp Smith, Commentary, 94“7, and Al-Azm, The Origins of Kant™s Argu-
ments in the Antinomies, 11. Allison locates the original terminology in Erdmann, Re¬‚exionen,
393. See Allison, Kant™s Transcendental Idealism, 43 and 338n26.
14 Melnick™s main discussion of the First Antinomy is in chapter 2 of Space, Time, and Thought
in Kant, 329“53.
Transcendental illusion II 235
individuated, a transcendental realist could not conceive the totality
of either temporal or spatial parts of the world as an analytic whole.
Analytic wholes are those in which the whole is prior to the parts. This
means the parts have no real existence in themselves independently of
the whole, but come into existence (as parts) only as constructed by the
marking process. The Aesthetic showed that our space and time are
analytic wholes. Now consider what it would mean to claim that
the world is an analytic whole. For temporal states, the existence of
each state of the world would depend on the existence of all the oth-
ers, entailing that the present depends also on the future as well as the
past. For spatial parts this means, similarly, that no particle of matter
could exist without the existence of all particles of matter. It is hard
to see how a realist could defend such a conception of the world.
Melnick argues that transcendental realists must accept Kant™s view
that the world-whole in space and time is a synthetic whole, precisely
because they represent spatiotemporal things as self-subsistent “by
representing them as all there to be met with by our procedures.”15
Because the world is there to be encountered by the subject, the parts
must be thought as given independently of the constructive process. Since
humans do not intuit the world as a whole, Kant seems justi¬ed in
claiming that the idea of the world-whole arises by synthesis of the
parts. Now since all synthesis is successive for humans, whether the
parts exist successively or simultaneously, the thought of its compo-
sition requires a successive synthesis. And since representing an in¬-
nite series as completed is impossible, “this completion, hence also
its concept, is impossible” (A432/B460). Thus the thesis argument
concludes: “Therefore an in¬nite given magnitude, and hence also
an in¬nite world (regarding either the past series or extension), is
impossible” (emphasis mine; A430/B458).
This reading highlights both strengths and weaknesses of Kant™s
argument. The weak points are the theoretical assumptions that the
idea of the whole must arise by synthesis from the parts, and that
human synthetic thought is successive. On the other hand, this inter-
pretation shows the argument to escape some standard criticisms.
Allison discusses several common objections by different commen-
tators.16 Bertrand Russell raises two of them, one to Kant™s concept

15 Space, Time, and Thought in Kant, 322.
16 See Allison, Kant™s Transcendental Idealism, 40“5.
Transcendental illusion II
of in¬nity, and the other to the argument itself. First Russell objects
to introducing the notion of synthesis in the idea of in¬nity, since
by the Cantorian mathematical concept of in¬nity, “classes which are
in¬nite are given all at once by the de¬ning property of their mem-
bers.”17 This objection is misguided, however, for as we have seen,
Kant claims not that the concept of the in¬nite requires synthesis, but
rather that thinking of an in¬nite whole made of given parts requires
a synthesis. Kant has no objection to the mathematical concept of an
in¬nite set of members.
Strawson shares Russell™s second objection to the idea that an in¬-
nite series cannot be completed. Russell says, “all that [Kant] has even
conceivably a right to say is that it cannot be completed in a ¬nite
time. Thus what he really proves is, at most, that if the world had
no beginning, it must have already existed for an in¬nite time.”18
That is, either the world begins in time or it has existed in¬nitely.
If Kant is entitled to assume only that an in¬nite series cannot be
completed in a ¬nite time, then the argument proves, contrary to
its purpose, that the ¬rst alternative is false, and that the world is
in¬nite in time. Both Russell and Strawson apparently assume that it
makes sense to say that an in¬nite series can be “completed” in an
in¬nite time. But if the idea of an in¬nite is the idea of a number
greater than any ¬nite number, then, regardless of the time allot-
ted, it appears no such series could be thought (successively) as a
completed whole.19 Both objections miss the target as interpreted

Antithesis: “The world has no beginning and no bounds in space, but
is in¬nite with regard to both time and space” (A427/B455). Again
Kant offers separate proofs for time and space. The ¬rst paragraph
contains the argument for time:

17 Russell, Our Knowledge of the External World, 123.
18 Russell, Our Knowledge of the External World, 123, and Strawson, The Bounds of Sense, 176.
19 The same response can be made to a related objection by G. E. Moore and Jonathan Bennett,
that Kant wrongly inferred the impossibility of an in¬nite series with one bound from the
impossibility of an in¬nite series bounded at both ends. Allison points out the irrelevance,
since Kant “does not claim that a series cannot be in¬nite if it has one end . . . His point
is rather that since, as in¬nite, the series has only one end, it cannot constitute a totality”
(44). The issue is whether it is possible to think a totality as composed of an in¬nity of parts
through a successive synthesis.
Transcendental illusion II 237
1. Suppose the world has a beginning in (in¬nite) time.
2. “Since the beginning is an existence preceded by a time in which
the thing is not, there must be a preceding time in which the world
was not, i.e., an empty time.”
3. But time is homogeneous: no part of time has “any distinguishing
condition of its existence rather than its non-existence.”
4. By 3, “no arising of any sort of thing is possible in an empty time.”
5. “Thus many series of things may begin in the world, but the
world itself cannot have any beginning, and so in past time it is
In this case, the antithesis argument for space appears to differ from
the argument for time:
1. Assume the opposite, “namely that the world is ¬nite and bounded
in space; then it exists in an empty space, which is not bounded.”
2. By 1, there would be not only relations “between things in space,
but also a relation of things to space.”
3. But “the world is an absolute whole, besides which there is encoun-
tered no object of intuition,” and therefore nothing else to which
the world could be related.
4. Hence “the relation of the world to empty space would be a relation
of the world to no object. Such a relation, however, and hence
also the boundedness of the world by empty space, is nothing.”
5. Therefore “the world is not bounded at all in space, i.e., in its
extension it is in¬nite.”
These proofs are aimed against the view that the world is ¬nite in
absolute time and absolute space. On this conception, the world
would begin at a time preceded by (in¬nite) empty time, and be
surrounded by (in¬nite) empty space. Both proofs argue that these
are incoherent conceptions. The ¬rst proof explicitly invokes the
Principle of Suf¬cient Reason, claiming that there is no suf¬cient
basis in time for the world to begin at one particular moment rather
than another. Clearly this same argument applies to space: absolute
space provides no suf¬cient basis for locating the world in one region
rather than another. But if the world were temporally and spatially
¬nite, then it would occupy determinate regions of absolute time and
absolute space.
Transcendental illusion II
The second argument makes a different point concerning space: if
the world is ¬nite in absolute space, then it has what Melnick calls
“multiple situatability.”20 That is, if it is bounded at S1 , then it has the
determinate relation of being 10 feet away from some empty space
S2 distinct from S1 , and so on for all its relations to all other empty
spaces. Because space is homogeneous, there is no possible way to
think the difference in that situation, and one in which the world is
shifted exactly 20 feet away from S2 . Yet for the absolutist they must be
distinct. Kant puts the point in terms of the correlates of the relation:
because the world encompasses all that is real in space, there is nothing
real against the world on which to ground its determinate relation
to space. “Such a relation, however, and hence also the boundedness
of the world by empty space, is nothing” (A429/B457). Therefore,
“the world is not bounded at all in space, i.e., in its extension it is
in¬nite.” If these determinate locations are not thinkable, then the
realist conception of a ¬nite world is incoherent.
In his remark on the antithesis Kant considers an alternative
¬nitism, based on a relational theory of space and time. This is the
Leibnizian view discussed earlier in chapter 3, according to which
space and time are not independent of the real, but are derived from
the relations among real things. As Kant explains, a relationist must
think of a ¬nite world abstracted from spatial and temporal limits,
since the boundaries of the world precede its “location” in space and
time: “instead of a ¬rst beginning . . . one thinks of an existence in
general that presupposes no other condition in the world, rather
than the boundary of extension one thinks of the limits of the world-
whole, and thus one gets time and space out of the way” (A433/B461).
In thinking the limits of the world as non-temporal and non-spatial,
the relationist must think “surreptitiously of who knows what intelli-
gible world in place of a world of sense.” But cosmology concerns the
nature of appearances in space and time. Thus a mundus intelligibilis
“is nothing but the concept of a world in general, and in regard to
which, consequently, no synthetic proposition at all, whether af¬r-
mative or negative, is possible” (A433/B461).
These antithesis arguments, based on the Principle of Suf¬cient
Reason, appear stronger than the thesis arguments. First, both

20 Melnick discusses the antithesis of the First Antinomy at Space, Time, and Thought in Kant,
329“44; the discussion of “multiple situatability” begins at 335.
Transcendental illusion II 239
Leibniz and Clarke accept this principle.21 And second, since the
unconditioned is the set of conditions jointly necessary and suf¬cient
for the given, accepting the demand of reason implicitly commits one
to some version of the Principle of Suf¬cient Reason. The proofs also
infer the impossibility of a bounded world from the impossibility of
thinking its location in absolute space and time. Without some logi-
cal basis for giving the world a determinate location in absolute space
and time, the realist would be hard pressed to reject the antithesis

B. The Second Antinomy: the nature of substance
Thesis: “Every composite substance in the world consists of simple
parts, and nothing exists anywhere except the simple or what is com-
posed of simples” (A434“6/B462“4).
The ¬rst paragraph contains the argument for the thesis:
1. Assume the opposite, that “composite substances do not consist of
simple parts.”
2. By 1, “if all composition is removed in thought, no composite part,
and (since there are no simple parts) no simple part, thus nothing
at all would be left over.”
3. If nothing at all would be left over, “no substance would be given.”
4. Implied: substance is given.
5. Therefore, either (a) “it is impossible to remove all composition in
thought” or (b) “after its removal something must be left over that
subsists without any composition, i.e., the simple.”
6. For substances, “composition is only a contingent relation, apart
from which, as beings persisting by themselves, they must subsist”
7. Therefore for substances it must be possible to remove all compo-
sition in thought: “the composite would once again not consist of
8. By 6 and 7, (a) is impossible.
9. By 5 and 8 it follows that “what is a substantial composite in the
world consists of simple parts.”
In the remark on this argument Kant points out that the conclusion
applies neither to space and time, nor to accidents of substances. First,
21 For Leibniz and Clarke, see Al-Azm, The Origins of Kant™s Arguments in the Antinomies, 30“5.
Transcendental illusion II
space and time are not substances. Second, as the Aesthetic showed,
although they are composed of parts, the “parts are possible only in the
whole” (A438/B466). Here he classi¬es them as ideal as opposed to real
composites.22 For space and time “if I remove all composition from it,
then nothing, not even a point, might be left over; for a point is pos-
sible only as the boundary of a space (hence of a composite)” (A438“
40/B466“8). The conclusion also does not apply to states or accidents
of substances, since they “do not subsist by themselves” (A440/B468).
This argument is based on conceiving a substance as a self-
subsistent being and, I suspect, on the view that relations are based
on non-relational properties of things. Despite the various theories of
substance, substances were generally conceived as independent enti-
ties. This implies that where a being is composed of substances, the
existence of the composite depends on the existence of the parts. This
is Kant™s point in line 6, where he claims that for composite substances,
composition is a “contingent relation.” Now if the composition of
composite substances is only a contingent property, then it must be
possible “to remove all composition in thought,” that is, to think
the real in the composition independently of the composite. Once
all composition is abstracted away, according to the argument, all
that remains are non-composite or simple parts. Although not stated
explicitly, the argument would apply to both material and mental
The above reasoning is valid only if being “self-subsistent” rules out
all contingent properties. But this is not obvious. From the fact that a
composite of substances must be composed of self-subsistent elements,
it does not follow that these elements could not be composite in
nature. Even if composition is a contingent property, substances could
be irreducibly composite if one conceived of self-subsistent elements
(substances) as possessing contingent properties.
22 Grier points out that despite the similarity to the earlier distinction between analytic and
synthetic wholes, Kant reserves the term “whole” or totum for the world as a whole, which was
the subject of the First Antinomy. The term “composite” or compositum applies generally
to anything in appearance made up of parts. See Grier, Kant™s Doctrine of Transcendental
Illusion, 196.
23 Grier makes this point forcefully. She criticizes Al-Azm for reading the thesis and antithe-
sis arguments as using different notions of substance, the thesis concerning only mate-
rial substance and the antithesis substances generally. See Al-Azm, The Origins of Kant™s
Arguments in the Antinomies, 46ff, and Grier, Kant™s Doctrine of Transcendental Illusion,
Transcendental illusion II 241
A different defense is suggested by the second paragraph, where
Kant says it follows that “all things in the world are simple beings,
that composition is only an external state of these beings” (my empha-
sis). He continues: “reason must still think of them as the primary
subjects of all composition and hence think of them prior to it as
simple beings” (A436/B464). This passage appeals to the principle
of the reducibility of relations discussed in the Amphiboly. On this
view, all relations or “external determinations” are reducible to non-
relational or “inner determinations” of things. Although Kant claims
transcendental realists must endorse the reducibility of relations, he
denies that it applies to appearances in space and time. Reading the
thesis proof this way yields a valid argument. Given both that com-
position is a relation among parts and that self-subsistence implies
only non-relational properties, self-subsistent substances must be sim-
ples. A rational metaphysician could avoid the conclusion by deny-
ing either the reducibility of relations or that self-subsistent entities
have only necessary properties. As Grier points out, however, if one
accepts reason™s demand for the unconditioned, the idea of a com-
posite that is not reducible to ultimate, simple parts fails to achieve

Antithesis: “No composite thing in the world consists of simple parts,
and nowhere in it does there exist anything simple” (A435/B463). The
¬rst paragraph argues that composites are not composed of simple
parts by locating composites in space:
1. Assume the opposite: suppose substances are composed of simple
2. Because composition is an “external relation between substances,”
it is possible only in space.
3. By 2, the space occupied by a composite thing must have as many
parts as the thing occupying it.
4. But space is in¬nitely divisible and does not consist of simple parts.
5. Therefore by 3 and 4 every simple part of the composite must
occupy a space.
6. By virtue of occupying space, every simple part contains parts
external to one another.
24 See Grier, Kant™s Doctrine of Transcendental Illusion, 203.
Transcendental illusion II
7. Hence every simple part is a composite of real substances.
8. Because 7 is self-contradictory, substances cannot be composed of
simple parts.
By contrast, the proof in the second paragraph that there are no
simples anywhere does not depend on their relation to space, but
rather on the nature of experience in general:
1. Assume that the transcendental idea of the simple applied to
2. By 1, the empirical intuition of such an object would have to be
3. Such an intuition would contain “absolutely no manifold whose
elements are external to one another and bound into a unity”
4. Implied: spatial and temporal intuition by its nature contains a
manifold of external elements bound together in a unity.
5. But “this intuition is de¬nitely required for absolute simplicity.”
6. By 4 and 5, the simplicity of anything given in appearance “cannot
be inferred from any perception, whatever it might be.”
7. Therefore nothing simple is given in the world of sense “regarded
as the sum total of all possible experiences.”
Rather than arguing against the existence of indivisible parts, this
second proof claims only that experience could in principle offer
no evidence for their existence. Because appearances are given in
space and time, and because empirical intuition inherently contains a
manifold, intuition could offer no grounds for inferring the simplicity
of anything given in appearance. This argument is aimed against both
material and mental simples.
The ¬rst proof sides with the understanding in locating the world
of appearance in space (and time). Although the proof does not men-
tion matter per se, the ¬rst paragraph of the remark indicates that
the substances under consideration are bodies (A441/B469). As Grier
points out, it is a mistake to interpret the argument as relying on
Kant™s theory in the Aesthetic. It is true that the argument concludes
that there are no simple parts of matter, based on the fact that the
space matter occupies is in¬nitely divisible. But as we shall see, in
his resolution Kant rejects that inference. Unlike Kant, the antithe-
sis assumes that space is transcendentally real, and concludes that
substance in itself is in¬nitely divisible.
Transcendental illusion II 243
In line 2 the proof moves from the claim that composition is an
“external relation” to the claim that composite substances must exist
in space. But the claim in line 6, that the mathematical divisibility
of space entails the real divisibility of whatever occupies it, is not
obviously true. Al-Azm offers two possible defenses. One is to read
the argument as making the Leibnizian point that postulating simples
violates the Principle of Suf¬cient Reason. That is, if one admits that
the real occupying space contains a manifold of external parts, “it
would be simply arbitrary” to stop at a real thing that is indivisible.25
Since the text does not explicitly mention this principle, his second
suggestion looks more promising. According to it, line 6 assumes that
the hypothesized simple parts occupy space “in exactly the same sense
as the composite object itself is said to be in space.”26 It would follow
that all parts reached by division are “external” to each other in the
same sense that the substances making up the composite are external.
Suppose, for example, one explains the spatial extension of matter in
terms of impenetrability. To say that one half of a body occupies a
different space from the other half is to say that the parts bear this
impenetrability relation to one another. If every part of matter bears
this relation to every other part of matter, one can never arrive at a
non-divisible “simple” that stands in no impenetrability relation to
another space-occupying part.27
As the last paragraph of the remark indicates, the second proof is
also aimed against the view that the thinking self is a simple substance.
As we saw in the Paralogisms, Kant™s argument against the simple soul
depends on his idealistic principle that the thinking thing in itself is
not given in experience. Here he wants to show, independently of
transcendental idealism, that philosophers who claim to have imme-
diate awareness of a simple self are mistaken. Given that conclusion,
it is puzzling to ¬nd Kant claiming that nothing in inner sense “could
prove a manifold of elements external to one another, and hence real

25 Al-Azm, The Origins of Kant™s Arguments in the Antinomies, 63“4.
26 Al-Azm, The Origins of Kant™s Arguments in the Antinomies, 61.
27 In the MFNS of 1786, Kant argues for a dynamical theory of matter as composed of centers
of repulsive force. As Michael Friedman explains in his introduction to the translation,
“matter is explicitly taken to be continuous or in¬nitely divisible, and material substance, in
particular, is now characterized precisely by the impossibility of elementary monadic simple
elements.” Theoretical Philosophy after 1781, 174. Contrary to the antithesis argument of the
Second Antinomy, Kant™s theory depends on his transcendental idealism, and the view that
matter is only appearance and not a thing in itself.
Transcendental illusion II
composition” (A443/B471; my emphasis). I think his point, however,
is that the data of inner sense cannot be used either for or against
the existence of a simple self. As the stated proof points out, aware-
ness through inner sense is inherently complex and therefore cannot
support the simplicity of mental substance. On the other hand, the
manifold of inner sense cannot prove anything about the subject “con-
sidered externally, as an object of intuition” (A443/B471). In other
words, any conclusion about the real nature of mental substance must
be based on its “external” existence in relation to other things, and
not merely on inner intuition. The point of the second proof is to
refute claims that empirical intuition could support the simplicity of
mental or material substance.

C. The Third Antinomy: freedom and determinism
Thesis: “Causality in accordance with laws of nature is not the only
one from which all the appearances of the world can be derived. It is
also necessary to assume another causality through freedom in order
to explain them” (A444/B472). The premises are contained in the
¬rst paragraph, A444“6/B472“4, and the conclusion is spelled out in
the second:
1. Assume the opposite, that there is only causality “in accordance
with laws of nature.”
2. By 1, “everything that happens presupposes a previous state, upon
which it follows without exception according to a rule.”
3. By 2, this applies to every state of the world-series, and so on ad
4. By 3, there is no ¬rst beginning, and thus “no completeness of the
series on the side of the causes descending one from another.”
5. But “the law of nature consists just in this, that nothing happens
without a cause suf¬ciently determined a priori.”
6. Thus the assumption that “all causality is possible only in accor-
dance with laws of nature . . . contradicts itself.”
7. Therefore there must be a causality “through which something hap-
pens without its cause being further determined by another pre-
vious cause, i.e., an absolute causal spontaneity beginning from
itself . . . hence transcendental freedom.”
Transcendental illusion II 245
As the remark points out, the conclusion establishes that there must
be a ¬rst cause that has the spontaneous power to begin the series
of world states. But it also opens up the possibility that there are
spontaneous causes operating within the world-series:
because the faculty of beginning a series in time entirely on its own is thereby
proved . . . now we are permitted also to allow that in the course of the world
different series may begin on their own as far as their causality is concerned,
and to ascribe to the substances in those series the faculty of acting from
freedom. (A450/B478)

The thesis argument aims to prove the existence of transcendental
freedom, or a cause having the power to initiate an event sponta-
neously, without prior determination. But the idea of transcendental
freedom makes possible the concept of free will or practical free-
dom, the power of a rational agent to choose independently of sen-
suous determinations (A533“4/B561“2). The different series within
the world that “begin on their own as far as their causality is con-
cerned” would include consequences ensuing from such free choices.
Since such actions have “natural consequences to in¬nity, there begins
an absolutely new series, even though as far as time is concerned this
occurrence is only the continuation of a previous series” (A450/B478).
According to this conception, the events initiated by the original, tran-
scendentally free occurrence could be either causally determined or
The key to the argument is line 5, that causal determinism requires
that “nothing happens without a cause suf¬ciently determined a pri-
ori.” Clearly this expresses the transcendental demand that causal
explanations terminate in a complete set of suf¬cient conditions for
the given. Only on this assumption can one avoid the possibility of an
in¬nite regress of causal states. What this does, of course, is to turn the
original motive for causal explanations against itself, exploiting the
“inherent tension” in the demands of reason. While admitting that
transcendental freedom cannot be explained, its proponent claims
that this is true of causal connections themselves: “with causality in
accordance with natural laws . . . we do not in any way comprehend
how it is possible for one existence to be posited through another exis-
tence” (A448/B476). So the thesis emphasizes the demand for closure
in the causal series, and concludes that a suf¬cient account requires
Transcendental illusion II
a cause not subject to deterministic connections. When challenged
to explain that cause, the proponent claims to be no worse off than
the determinist, since, as Hume demonstrated, there are no a priori
explanations for causal connections.

Antithesis: “There is no freedom, but everything in the world happens
solely in accordance with laws of nature” (A445/B473). This argument
occurs in a very compressed form in the ¬rst paragraph:
1. Assume the opposite, that there is an uncaused beginning to the
causal series of appearances.
2. By 1, there would exist a ¬rst state S1 , with the power to begin abso-
lutely another state S2 , “and hence also a series of its consequences,”
S3, S4, and so on.
3. By 2, the “determination of this spontaneity itself,” the causality of
S1 “will begin absolutely, so that nothing precedes it through which
this occurring action is determined in accordance with constant
4. But “a dynamically ¬rst beginning of action presupposes a state
that has no causal connection at all with the cause of the previous
one, i.e., in no way follows from it.”
5. Therefore “transcendental freedom is contrary to the causal law,
and is a combination between the successive states of effective
causes in accordance with which no unity of experience is pos-
sible . . . and hence is an empty thought entity.” (A445“7/B473“5)
Unlike Strawson, who sees the argument simply as endorsing the
universal principle of causality, Al-Azm notes the subtle way it explores
the notion of causal imputation. The point is to show that the idea
of a spontaneously acting cause that initiates a causally determined
series is incoherent. But the argument is not easy to make out. Line
3 states that, by de¬nition, the causal action of the transcendentally
free cause in state S1 is not determined by its antecedent states nor is
it governed by constant laws. The confusion arises with line 4, which
appears to repeat the point in line 3. Al-Azm™s account suggests the
following reconstruction.28 Suppose the dynamical ¬rst beginning of

28 I have simpli¬ed his presentation; see Al-Azm, The Origins of Kant™s Arguments in the Anti-
nomies, 103“5.
Transcendental illusion II 247
action mentioned in line 4 refers to the causality of S2 rather than S1 .
Kant™s point, then, is that the deterministic causality of the series S2 ,
S3 , S4 , . . . is not imputable to its antecedent condition S1 , since S1 acts
spontaneously. Thus there are really two causal ¬rst beginnings here,
the spontaneous action of S1 on S2 , and the deterministic action of
S2 on S3 . Because S1 does not act causally by constant, deterministic
laws, it cannot cause S2 to act causally by constant, deterministic laws.
Thus the concept of a spontaneous cause initiating a deterministic
causal series is incoherent.
As Al-Azm sees it, the argument raises a question about the relation
between the spontaneous act of origination “and the agent presum-
ably ˜responsible™ for that act.”29 This suggests a tactic used by oppo-
nents of free will, who rejected the idea of undetermined choice pre-
cisely because it con¬‚icts with moral responsibility. For a spontaneous,
undetermined choice would be one not connected to antecedent con-
ditions, including the agent™s character. In that case it could not be said
to be the action of the agent. The antithesis argument here makes
the parallel point that to attribute causality to a state presupposes
that it follows from the nature (and antecedents) of the state to act in
that manner. Thus spontaneous causality cannot provide a suf¬cient
explanation of a series of causally determined states.
In the remarks on the antithesis, Kant relates the Third Antinomy
to the First Antinomy. At A449/B477 he points out that the success
of the thesis argument for a ¬rst dynamical state depends on the con-
clusion that there is a ¬rst temporal state of the world. If one admits
that substances have always existed, then “there is no dif¬culty in also
assuming that the change of their states, i.e., the series of their alter-
ations, has always existed, and hence that no ¬rst beginning, whether
mathematical or dynamical, need be sought.” Moreover, the second
paragraph picks up the point stated in the proof at A447/B475, that
admitting transcendental freedom destroys the unity of experience:
For alongside such a lawless faculty of freedom, nature could hardly be
thought any longer, because the laws of the latter would be ceaselessly mod-
i¬ed by the former, and this would render the play of appearances, which in
accordance with mere nature would be regular and uniform, confused and
disconnected. (A451/B479)

29 Al-Azm, The Origins of Kant™s Arguments in the Antinomies, 105.
Transcendental illusion II
In other words, the existence of transcendental freedom within
appearances would make nature indeterministic. Not only would
it be impossible to predict the consequences of events, but, as Kant
argued in the Second Analogy, it would be impossible to distinguish
between an event, an objective succession of states, and a subjective
succession of perceptions. Clearly the antithesis position sides with
the principles of the understanding.
On the above reading, the antithesis argument also appeals to the
Principle of Suf¬cient Reason. Only instead of emphasizing suf-
¬ciency in a complete set of conditions, the latter emphasizes a
suf¬cient explanation of deterministic causal connections. Corre-
sponding to each strength is a weakness: the thesis can explain neither
the source of spontaneity nor the relation between spontaneity and
determinism; the antithesis cannot give closure to the causal series.
These corresponding strengths and weaknesses illustrate both the ten-
sion in applying the Principle of Suf¬cient Reason, and the con¬‚ict
between reason and the understanding.

D. The Fourth Antinomy: contingency and necessity
Thesis: “To the world there belongs something that, either as a part
of it or as its cause, is an absolutely necessary being” (A452/B480).
The thesis states that something absolutely necessary exists within
the world. The proof consists of two parts, the ¬rst arguing for an
absolutely necessary being, and the second arguing that this being
must exist in the world. The ¬rst part consists of these steps:
1. The sensible world as the whole of appearances contains a series
of alterations.
2. “Every alteration, however, stands under its condition, which pre-
cedes it in time, and under which it is necessary” (A452/B480).
3. Every given conditioned presupposes a complete series up to the
unconditioned, “which alone is absolutely necessary.”
4. “Thus there must exist something absolutely necessary, if an alter-
ation exists as its consequence.”
The second part argues that this necessary being cannot be outside
the world of appearances:
5. Assume the opposite, that the absolutely necessary being is outside
the world of sense.
Transcendental illusion II 249
6. By 5, the series of alterations in the world “would derive from it,
without this necessary cause itself belonging to the world of sense”
7. But “the beginning of a time-series can be determined only through
what precedes it in time.”
8. By 7, “the supreme condition of the beginning of a series of
changes” must exist in the time before the series comes into exis-
9. By 8, the absolutely necessary cause of the series “belongs to time,
hence to appearance (in which alone time is possible, as its form);
consequently it cannot be thought as detached from the world of
Although the argument seems straightforward, it turns out to be
more complicated than it appears on the surface. The main issue is
how it differs from the Third Antinomy argument for a transcen-
dentally free cause. Commentators such as Kemp Smith and Bennett
claim this proof is redundant, since the necessary being argued for
here is just the ¬rst, uncaused cause at issue in the preceding Anti-
nomy.30 Although the second part suggests this reading, Grier argues
persuasively that the two arguments have different purposes.31 First,
there is the obvious point that the two proofs involve different cat-
egories, the Third Antinomy causality, and the fourth the modal
concepts of necessity and contingency. As Kant explains in discussing
the ontological argument in the next section, the nominal de¬nition
of an absolutely necessary being is one whose non-being is impossible
(A592“3/B620“1). Notice that on this de¬nition, even if the necessary
being exists in time, there could be no time at which it did not exist.
That alone would rule out the idea that it is merely the ¬rst tempo-
ral state of a causal series. The natural application of this notion to
appearances would be to substance: rather than taking the necessary

30 See Kemp Smith, Commentary, 495, and Bennett, Kant™s Dialectic, 241.
31 See Grier, Kant™s Doctrine of Transcendental Illusion, 219“27. I am not as convinced, however,
by her claim that the Third Antinomy also does not involve the notion of a ¬rst causal
state in time. She cites as evidence Kant™s claim in the remark on the thesis that “here we
are talking of an absolute beginning not, as far as time is concerned, but as far as causality
is concerned” (A450/B478). See Grier, 220“1. She is right that the issue there is whether
deterministic causality must be conditioned by transcendental freedom. As we saw above,
however, at least the antithesis position takes the thesis argument to presuppose a ¬rst causal
state in time. Whether she is right about the Third Antinomy, her point seems stronger with
respect to the Fourth.
Transcendental illusion II
being as a ¬rst state of the series, it makes more sense to take it as some-
thing substantial existing permanently in time. In fact, in the solution
Kant says, “Here we deal not with unconditioned causality, but with
the unconditioned existence of the substance itself” (A559/B587).
Grier also points out that the references to cause in the Fourth Anti-
nomy can be construed in terms of immanent rather than transitive
causation. As Spinoza distinguished them, transitive causation occurs
between really distinct things, for example in a collision in which one
body causes another to move. An immanent cause, by contrast, is a
ground of something, inseparable from its effect. The numbers 1 and
2 can be seen as the “immanent causes” of the number 4, for exam-
ple, insofar as they are contained in it.32 Thus Grier maintains the
Fourth Antinomy treats necessary existence as the immanent cause of
all contingent existence in appearances rather than a temporally ¬rst,
transitive cause.
The ¬rst part argues directly from the contingency of appearances
to an absolutely necessary being. The description of appearances as a
series of alterations establishes their contingency, since, as line 2 spells
out, an alteration is an event necessitated by a temporally prior con-
dition. Although this presupposes a causal account, what is relevant
to the argument is the contingency. Moreover, we should note that
the necessity obtaining between empirical causes and effects is relative
rather than absolute. That is, if an empirical state follows necessar-
ily from a prior state, then its existence is not absolutely necessary
in Kant™s sense. Line 3 expresses the demand of reason for totality
in the conditions, which can be satis¬ed only by something whose
existence is absolutely necessary. This conclusion is stated in line 4,
which describes alteration as its consequence. On Grier™s reading, the
term “consequence” should be understood as an ontological rather
than a temporal effect.
The second part of the proof argues that this absolutely necessary
ground must belong to the world “either as a part of it or as its cause.”
Here Kant returns to the reductio method, and derives a contradiction
from the idea that the necessary being is outside appearances. The key
is in lines 7 and 8, which argue that the “beginning” of a time-series

32 See Spinoza, Ethics, part I, proposition 18, in The Ethics and Selected Letters, 46 and 25 of
Shisley™s introduction.
Transcendental illusion II 251
must itself exist in the time prior to the series. As I argued above,
Kant intends to apply this to substance rather than to a (temporary)
¬rst state of the series. The issue is whether it has to exist temporally
or can be conceived of as existing outside time. Clearly, if it exists at
all in time, then it exists at all times. Kant addresses the atemporal
version of “grounding” in the third paragraph of the remark. If the
“condition must be taken in just the same signi¬cance as it has . . .
in the series,” and the series takes place in time, then “the necessary
being must be regarded as the supreme member of the world-series”
(A457“8/B485“6). Unfortunately it is not obvious that the relation
among appearances is relevant if the issue is whether the contingency
of the entire series requires an absolutely necessary ground.
The last two paragraphs of the remark respond to this objection.
Here Kant argues that a shift from a cosmological to an intelligi-
ble necessary being confuses empirical and intelligible contingency
(A458/B486). By empirical contingency he means “that the new state
could not at all have occurred on its own, without a cause” in the pre-
vious time (A460/B488). An intelligible contingency is one “whose
contradictory opposite is possible” (A458/B486). A body changing
state from motion to rest is an example of empirical contingency,
since motion at one time does not contradict rest at another time
(A460/B488). The contradictory opposite of a state would require
“that at the very time when the previous state was, its opposite could
have been there in place of it.” In other words, if the absolutely neces-
sary being were outside time, it could only ground logical contingency.
Because the contingency here is empirical, the necessary being must
be in time, in the world of appearances.
Antithesis: “There is no absolutely necessary being existing anywhere,
either in the world or outside the world as its cause” (A453/B481). The
antithesis explicitly contradicts not only the thesis, but also the view
rejected above, that there is an absolutely necessary being outside
the world of appearances. The proof devotes a paragraph to each
alternative. The argument against the thesis is this:
1. Assume the opposite, “that either the world itself is a necessary
being or that there is such a being in it.”
2. By 1, in the series of alterations, either (a) “there would be a
beginning that is unconditionally necessary, and hence without
Transcendental illusion II
a cause,” or (b) “the series itself would be without any beginning,
and although contingent . . . it would nevertheless be absolutely
necessary and unconditioned as a whole.”
3. Alternative (a) con¬‚icts with the “law of the determination of all
appearances in time,” and so is not possible.
4. Alternative (b) is self-contradictory “because the existence of a
multiplicity cannot be necessary if no single part of it possesses an
existence necessary in itself.”
5. Therefore the original assumption is not possible.
The next paragraph argues that an absolutely necessary cause cannot
exist outside the world, as follows:
6. Assume “there were an absolutely necessary cause of the world
outside the world” (A453“5/B481“3).
7. By 6, “this cause, as the supreme member in the series of causes
of alterations in the world, would ¬rst begin these changes and
their series” (A455/B483).
8. But its action would “begin to act then, and its causality would
belong in time, and for this very reason in the sum total of appear-
ances, i.e., in the world.”
9. Therefore, “this cause would not be outside the world, which
contradicts what was presupposed.”
10. Therefore, “neither in the world nor outside it (yet in causal
connection with it) is there any absolutely necessary being.”
A footnote to line 7 distinguishes two senses of “begin,” one active
(transitive), meaning to initiate, and the other passive, referring to a
temporal commencing. Kant says, “I infer here from the former to the
latter.” So the inference from line 7 to line 8 appears to incorporate
the argument from the thesis that by virtue of initiating a series of
appearances in time, the necessary cause would also have to be in
time. Here the point is used ultimately against the existence of an
absolutely necessary being.
The ¬rst stage of the proof appears straightforwardly to follow the
logic of the Principle of Suf¬cient Reason. Again, siding with the
understanding, the antithesis argument ¬rst rules out an absolutely
necessary being as part of the world, since its existence contradicts the
principle of causality, which requires every contingency to be condi-
tioned by a further contingency. The more interesting argument is
Transcendental illusion II 253
against the second alternative, that the entire world of appearances is
absolutely necessary. The proof rejects this possibility on the grounds
that the idea of a necessary series composed entirely of contingent
parts is incoherent. This objection thus applies the Principle of Suf¬-
cient Reason in a direction opposed to the thesis argument, claiming
that the contingency of the parts does not provide a suf¬cient basis
for the necessity of the whole.

3. ka nt™s res olu tions a nd t ra nsc enden ta l
idea li s m
The remainder of the chapter falls into three parts. Sections 3 through
5 contain general remarks about the arguments. In sections 6 and
7 Kant discusses their relation to transcendental idealism. He then
presents his solution in section 8, and applies it in detail to each argu-
ment in section 9. The newest material here concerns the Third Anti-
nomy debate over determinism and transcendental freedom. Kant
explains at length how human actions can be subject to causal laws as
appearances, and also attributed to free will as their intelligible cause.
This is important as a preamble to his moral theory, presented in
the Groundwork of the Metaphysics of Morals and Critique of Practical
Earlier I discussed Kant™s claim that the thesis positions repre-
sent the “dogmatism” of pure reason, and the antithesis positions
the “pure empiricism” of the understanding (A466/B494). Following
that description in the third section, Kant evaluates the advantages
and disadvantages of each position. The dogmatic theses have the
advantage of supporting practical interests: Kant describes them as
“so many cornerstones of morality and religion” (A466/B494). By
contrast, the antithesis “robs us of all these supports,” and as a con-
sequence, “moral ideas and principles lose all validity” (A468/B496).
On the other hand, in rejecting reason™s demand for completion of the
series, the antithesis arguments promote the interests of speculative
reason by making continuing inquiry possible. By contrast, the dog-
matist introduces “ideas with whose objects it has no acquaintance
because, as thought-entities, they can never be given” (A469/B497).
Dogmatism thereby abandons natural inquiry, “certain that it can
never be refuted by facts of nature because it is not bound by their
testimony.” Given the nature of these con¬‚icts, in the absence of
Transcendental illusion II
practical and speculative interests, one “would be in a state of cease-
less vacillation” (A475/B504), one day persuaded by the thesis, the
next by the antithesis.
In the fourth section Kant claims that because the con¬‚icts arise
from the inherent tension within reason, they are all resolvable by rea-
son. The fact that the object is the empirical cosmos implies that the
resolution will derive from the empirical synthesis on which the tran-
scendent idea is based (A479/B507). The ¬fth section gives a “skeptical
representation” of the con¬‚icts, describing them as between ideas that
are either too big or too small for the concept of the understanding.
For the ¬rst three Antinomies, the thesis conclusions are “too small,”
since they close off the series. The opposing antithesis conclusions
asserting the in¬nity of the mathematical and dynamical series are
“too big” for the concepts of the understanding. The pattern breaks
with the Fourth Antinomy, where Kant says the thesis idea of an
absolutely necessary being is “too big for your empirical concept,”
while the antithesis position that all existence is contingent “is too
small for your concept” (A489/B517).
The sixth and seventh sections relate the con¬‚ict to transcendental
idealism. Section 6 distinguishes transcendental idealism from both
transcendental realism and empirical idealism. For transcendental
idealism, “objects of experience are never given in themselves, but
only in experience” (A492“3/B521). But “experience” means possible
rather than actual perception:
That there could be inhabitants of the moon, even though no human being
has ever perceived them, must of course be admitted; but this means only
that in the possible progress of experience we could encounter them; for
everything is actual that stands in one context with a perception in accor-
dance with the laws of the empirical progression. (A493/B521)

The empirical idealist, to the contrary, tries to reduce all objects to
collections of actual perceptions, and has dif¬culty accounting for
possible perceptions.
Section 7 then sketches the general form of Kant™s resolution:
although the conclusions of the arguments appear contradictory, they
are not. Instead, the opposition is “dialectical” as opposed to “analyti-
cal” (A504/B532). At A503/B532 Kant cites as examples the judgments,
“every body smells good” and “every body smells not good,” which
Transcendental illusion II 255
are not contradictories, since they both assume that every body has
some smell. If, however, there are bodies that lack an aroma, then the
propositions are contraries, since they can both be false. Similarly,
all the Antinomies presuppose that the world as the whole series of
appearances is a thing in itself. If this were true, then the conclusions
would contradict each other, with one true and the other false. But
if appearances are not things in themselves, then the world “does not
exist at all (independently of the regressive series of my representa-
tions)” and “by itself it is not to be met with at all” (A505/B533). And
at A506“7/B534“5 Kant offers this dilemma to show how the ¬rst
two Antinomies support transcendental idealism: “If the world is a
whole existing in itself, then it is either ¬nite or in¬nite. Now the ¬rst
as well as the second alternative is false . . . Thus it is also false that
the world (the sum total of appearances) is a whole existing in itself.”
As we shall see below, for the dynamical Antinomies, Kant offers a
different resolution.
The eighth section explains the principles of Kant™s resolution.
He recalls that reason™s idea of the unconditioned is only regulative,
supplying a maxim for inquiry, rather than constitutive, making a
substantive claim about the object: “Thus the principle of reason
is only a rule, prescribing a regress in the series of conditions for
given appearances, in which regress it is never allowed to stop with an
absolutely unconditioned” (A509/B537). Thus the principle “cannot
say what the object is, but only how the empirical regress is to be
instituted” (A510/B538). The rest of the section distinguishes between
two sorts of empirical regress, one to in¬nity (in in¬nitum), the other
extending indeterminately (in inde¬nitum).
A regress to in¬nity applies where the whole is empirically given.
For example, in dividing a body (or a line segment), the process can
go on to in¬nity since the parts (conditions) are given with the whole.
Because here “an unconditioned (indivisible) member of this series
of conditions is never encountered . . . the division goes to in¬nity”
(A513/B541). Where one is given only a member and seeks to extend
the series, the regress is inde¬nite rather than in¬nite. For example,
in tracing someone™s ancestors, because the whole series is not given,
“this regress . . . goes to an indeterminate distance, searching for more
members for the given” (A523/B541). The rule for the in¬nite regress
is, “You ought never to stop extending it,” because one is assured that
Transcendental illusion II
there is always a further member given empirically with the whole. By
contrast, the rule for the inde¬nite regress is, “Extend it as far as you
want,” because no member can be given as absolutely unconditioned
(A511/BH539). In neither regress, however, is the series being given
“in¬nite in the object” (A514/B542). Since the objects of the regress
are only appearances, the conditions “ parts or further members “ are
given only in the regress. In sum, for appearances one cannot determine
how big the series of conditions is, either ¬nite or in¬nite, “for it is
nothing in itself.” Although we can know a priori that space and time
are in¬nite, there is no determinate answer to the question, is the
world in¬nite in space and time?
The particular solutions follow from this analysis. For the First
Antinomy, the question is whether the world is bounded by empty
time or space. Since experience is always of the conditioned “ i.e., an
empty space or time beyond the world is not a possible object of expe-
rience “ one could never encounter the boundary of the world. Like
the inquiry into one™s ancestors, the search for the conditions goes on
in inde¬nitum: one is not assured of encountering a further member
of the series, but neither can one assume an unconditioned mem-
ber. In consequence, as Kant puts it in a footnote: “This world-series
can thus be neither bigger nor smaller than the possible empirical
regress . . . And since this cannot yield a determinate in¬nite, nor
yet something determinately ¬nite . . . we can assume the magni-
tude of the world to be neither ¬nite nor in¬nite” (A518/B546). Thus
there is no determinate answer to the question: how big is the world?
In a second footnote Kant notes the difference between his position
and the antithesis view that the world is actually in¬nite in time and
space (A521/B549). For Kant, both the thesis and antithesis are false
of appearances.
The same reasoning applies to the Second Antinomy, concern-
ing the divisibility of the real. In this case, since bodies are given
in experience, the regress is in in¬nitum, meaning that one must
continue seeking the condition (parts) for every member encoun-
tered. But although one can never arrive at simples, neither is one
entitled to claim, with the antithesis, that the whole is composed
of an in¬nity of parts: although all the parts are contained in the
intuition, “the whole division is not contained in it; this division
consists only in . . . the regress itself, which ¬rst makes the series
Transcendental illusion II 257
actual” (A524/B552). Here there are also two cases, one for matter
as continuous quantity (quantum continuum), another for matter as
discrete (quantum discretum). In the ¬rst case matter is not articulated
into parts, and the division proceeds to in¬nity as it does for space.
In the second case matter is articulated, as in an organic body. Here,
“only experience can settle how far the organization in an articulated
body may go, and . . . such parts must nevertheless at least be within
a possible experience” (A527/B555). In general, however, the extent to
which appearances can be divided “is not a matter of experience”; it
is “a principle of reason never to take the empirical regress . . . to be
absolutely complete.” As Melnick explains, the transcendental realist
can apply the idea of in¬nity to a whole given of parts. For the tran-
scendental idealist, because an in¬nite series cannot be completed, the
idea of in¬nity applies only to the rule for seeking the condition.33
In concluding his account of the mathematical Antinomies, Kant
explains that they admit of a “both false” resolution because the con-
ditions are homogeneous with the conditioned. When investigating
the temporal and spatial bounds of the universe, or the parts of the
given whole, “none other than a sensible condition can enter, i.e., only
one that is itself a part of the series” (A530/B558). For the dynami-
cal Antinomies the matter is different, since “a synthesis of things
not homogeneous . . . must be at least admitted in the case of the
dynamical synthesis.” In these cases the dynamic series allows for an
intelligible condition that is not part of the series. In consequence,
although the dialectical arguments collapse, the rational proposi-
tions “may both be true” if their signi¬cance is restricted to either
things in themselves or appearances (A531“2/B559“60). As we shall
see, however, this resolution provides no support for transcendental
The resolution of the con¬‚ict between transcendental freedom and
causal determinism follows the “both true” pattern. First Kant empha-
sizes that the causal principle of the understanding necessarily applies
to appearances (A532/B560). The idea of transcendental freedom orig-
inates in reason, and represents the power to begin a state “from
itself, the causality of which does not in turn stand under another
cause determining it in time in accordance with the law of nature”

33 See Melnick, Space, Time, and Thought in Kant, especially 379“95.
Transcendental illusion II
(A533/B561). This is the basis of the idea of practical freedom or
free will. Morality and religion assume that human beings can deter-
mine themselves, independently of causal necessitation (A534/B562).
Whereas for the transcendental realist, causal determinism could not
coexist with transcendental freedom, this is possible for the transcen-
dental idealist. Kant then explains how “the very same effect that is
determined by nature” can also allow for freedom (A536/B564).
Before looking at the details of Kant™s solution, there are two issues
to address brie¬‚y. One concerns the relation between Kant™s views
of freedom here and in his ethical theory. Allison argues that in 1781
Kant had not yet developed the notion of autonomy central to his
moral theory. Thus the idea of free will here is the negative idea
of the agent resisting determination by sensible impulses. Not until
the Groundwork of the Metaphysics of Morals (1785) and the Critique of
Practical Reason (1788) did Kant conceive of free will as autonomy, the
faculty for giving the law to oneself.34 Accordingly he also changes his
stand on our knowledge of freedom. In the ¬rst Critique he claims only
that transcendental freedom is conceivable; the moral theory argues
that transcendental freedom can be deduced from the existence of
the moral law.
A second issue is whether the Dialectic account of the relation
between transcendental and practical freedom is inconsistent with
remarks in the later Canon of Pure Reason. As we have seen, the
Antinomies treat practical freedom as in some sense dependent on
transcendental freedom. By contrast, in the Canon Kant says whether,
in actions of practical freedom, “reason is not itself determined by fur-
ther in¬‚uences,” does not concern us in the practical sphere, since “we
ask of reason only a precept for conduct; it is rather a merely specula-
tive question, which we can set aside as long as our aim is directed to
action or omission” (A803/B831). Here he allows the possibility that
the spontaneity exhibited in free will might not be the absolute spon-
taneity of transcendental freedom. This implies that transcendental
freedom is not presupposed by practical freedom.
Allison thinks the apparent contradiction between the texts can
be dispelled. First he claims Kant takes the dependence of practical

34 See Allison, Kant™s Transcendental Idealism, chapter 15, especially 315“17. I am indebted to
Allison™s explanation of transcendental and practical freedom.
Transcendental illusion II 259
on transcendental freedom as conceptual rather than real: “it is this
transcendental idea of freedom on which the practical concept of
freedom is grounded” (A533/B561). Since confusing transcendental
ideas for ideas of objects involves transcendental illusion, Kant could
not consistently claim that the reality of practical freedom presupposes
the reality of transcendental freedom. This introduces the possibility
mentioned in the Canon, that practical freedom is not the absolute
spontaneity conceived in the idea of transcendental freedom.35 But as
Kant claims, the speculative basis of practical reason is not an issue
for morality.
The “both true” resolution of the Third Antinomy begins with
Kant reaf¬rming that “if all causality in the world of sense were
mere nature, then every occurrence would be determined in time,”
and so abolishing “transcendental freedom would also simultaneously
eliminate all practical freedom” (A534/B562). Moreover, if appear-
ances were things in themselves, then “freedom cannot be saved,” for
nature would be a “determining cause, suf¬cient in itself, of every
occurrence” (A536/B564). But because appearances are not things in
themselves, “they themselves must have grounds that are not appear-
ances.” Although these “intelligible” grounds are outside appearances,
they nevertheless give rise to effects in the series. “The effect can there-
fore be regarded as free in regard to its intelligible cause,” and yet the
result of necessary laws in regard to appearances (A537/B565). The
remainder of this section explains how transcendental freedom and
determinism can coexist by distinguishing the intelligible from the
empirical character of action.
At A538/B566 Kant de¬nes the intelligible as “that in an object of
sense which is not itself appearance.” The “clari¬cation” applies this
de¬nition to human intellectual faculties. Through the t.u.a., one rec-
ognizes that acts of understanding and reason “cannot be accounted
at all among impressions of sense.” In consequence, subjects iden-
tify themselves as partly phenomenal, and partly merely intelligible
(A546“7/B575). Human actions, then, can have both an empirical
and an intelligible character, where “character” refers to the “law of
its causality” (A539/B567). In its empirical character, as subject to
sensible conditions, the action is connected to other appearances “in

35 See Allison, Kant™s Transcendental Idealism, 315“19.
Transcendental illusion II
accordance with constant natural laws.” By virtue of its intelligible
character the action would not stand under temporal conditions,
for time is only a condition of appearances. Considered as an effect
of an intelligible cause, “no action would arise or perish,” and so,
not being part of the empirical series that makes its necessary, the
action would be free of all causal determination (A541/B569). Thus
intelligible causality “begins its effects in the sensible world from
itself, without its action beginning in it itself.” Kant concludes that
“freedom and nature . . . would both be found in the same actions,
simultaneously and without any contradiction, according to whether
one compares them with their intelligible or their sensible cause”
As Allison explains, this is a “compatibilism” in which the empirical
and intelligible characters represent alternative ways of explaining the
action.36 The empirical account views the action as emanating from an
agent™s desires, themselves understood as determined by physiological,
psychological, and sociological causes. In presupposing that an indi-
vidual™s character develops as an effect of these conditions, empirical
explanations accord with the causal principle. Explanations by intel-
ligible causes appeal to the agent™s reasons for acting rather than to
natural causes. This “causality of reason” is expressed through imper-
atives, both moral and non-moral: “The ought expresses a species
of necessity and a connection with grounds which does not occur
anywhere else in the whole of nature” (A547/B575). Whereas there is
no room in nature for the idea that something ought to exist, this
˜ought™ expresses an action whose ground “is nothing other than a
mere concept.” Explanations in terms of reasons assume that choices
are governed by rational principles relating the action to the agent™s
The question, of course, is how both types of causation can work
together. Kant believes that when we exercise practical reason, our
desires function as incentives rather than causes of action. As Allison
explains, for beings with free will, an incentive can determine an
action only “insofar as the agent incorporates that incentive into
his rule or maxim of action.”37 When an agent acts freely on a
desire, the action is based on a maxim licensing the action on that

36 37
Allison, Kant™s Transcendental Idealism, 325“9. Kant™s Transcendental Idealism, 327.
Transcendental illusion II 261
desire: “In circumstances C, it is permissible (or obligatory) to act
on my desire D.” According to this model of rational agency, desires
are effective only insofar as agents subsume them under rules they
endorse. The intelligible character of free action consists in this act of
Kant uses the example of the malicious liar to illustrate his view.
An explanation in terms of the person™s empirical character seeks its
sources in upbringing and environment as well as natural temper-
ament, that is, “the occasioning causes” (A554/B582). Nevertheless,
we hold the agent responsible: “This blame is grounded on the law
of reason, which regards reason as a cause” that, independently of
conditions, ought to have determined the person to act otherwise
(A555/B583). This is possible because “one regards the causality of
reason not as a mere concurrence with other causes, but as complete
in itself . . . [R]eason, regardless of all empirical conditions of the
deed, is fully free, and this deed is to be attributed entirely to its fail-
ure to act” (A555/B583, emphasis added). On this model, the empirical
and intelligible accounts of action do not con¬‚ict, since the former
is incomplete and subject to temporal conditions. Reason is atem-
poral because the “act of incorporation” is timeless. Kant says, “In
regard to the intelligible character . . . no before or after applies”
(A553/B581). That is, although the imperatives under which one acts
apply to temporal events, one™s adherence to them is not part of the
causal series.
It follows that in judging free actions, “we can get only as far as the
intelligible cause, but we cannot get beyond it . . . But why the intel-
ligible character gives us exactly these appearances and this empirical
character” cannot be explained (A557/B585). In keeping with tran-
scendental idealism, Kant claims that this resolution proves only the
possibility of practical or transcendental freedom in the noumenal
realm. He ends the section by noting that his resolution demonstrates
that freedom and causal necessity are compatible: “since in freedom
a relation is possible to conditions of a kind entirely different from
those in natural necessity, the law of the latter does not affect the
former; hence each is independent of the other” (A557/B585).

38 For a discussion of this view, see Allison, Kant™s Theory of Freedom, part I: “Freedom and
rational agency in the Critique of Pure Reason,” 11“82.
Transcendental illusion II
Compared to this discussion, Kant™s resolution of the Fourth Anti-
nomy is mercifully short, following the pattern of the above resolu-
tion. First he points out that because every member in the series of
appearances is contingent, there is no unconditioned or absolutely
necessary member anywhere (A559/B587). So if appearances were
things in themselves, then there could be no absolutely necessary
being as their condition. But because the dynamic regress can postu-
late a heterogeneous condition, one outside the spatiotemporal order,
it is possible for contingent appearances to be grounded in an abso-
lutely necessary intelligible being. This “both true” resolution differs
from that of the previous Antinomy since “in the case of freedom,
the thing itself as cause (substantia phaenomenon) would nevertheless
belong to the series of conditions, and only its causality would be
thought as intelligible” (A561/B589). An absolutely necessary being,
however, could not exist in the sensible world, although it could be
“the ground of the possibility of all these appearances” (A562/B590).
In his concluding remark Kant notes that this transcendental idea of
an absolutely necessary intelligible ground of all existence is the basis
of rational theology, the subject of the next chapter.

4 . su mm a ry
In the Antinomies Kant examines the arguments of rational cosmol-
ogy, those concerning the nature of the world considered as the sum
total of appearances. The four metaphysical disputes, following the
four categorical heads, debate whether the world is in¬nite in space
and time, whether matter is in¬nitely divisible, whether all events
are causally determined, and whether there is an absolutely necessary
existence. Kant™s analysis shows how each thesis position endorses the
demand of reason for the unconditioned, while its antithesis presup-
poses the principles of the understanding. Kant offers a “skeptical”
resolution of the disputes, arguing that in no case are the conclusions
true contradictories. These disputes are signi¬cant for providing indi-
rect support for transcendental idealism. This applies most clearly to
the ¬rst two, mathematical, Antinomies. If appearances were things in
themselves, either the thesis or antithesis would have to be true. Since
in the mathematical Antinomies both conclusions are false of appear-
ances, appearances cannot be things in themselves. For the last two,
Transcendental illusion II 263
dynamical, Antinomies, Kant offers a “both true” resolution, which
presupposes the truth of transcendental idealism. In these cases the
thesis is possibly true of things in themselves, with the antithesis true
of appearances. This analysis of the metaphysical disputes reinforces
the critical theory that the synthetic a priori principles of the under-
standing apply only to appearances, and not to things in themselves,
and thus exposes the illusion in attempting to take the regulative
demand of reason for constitutive concepts of objects.
ch a p t e r 10

Transcendental illusion III: rational theology

Kant has a complex attitude toward religion. One one hand he con-
sistently rejects religious belief based on superstition, fanaticism, and
anthropomorphism. He especially opposes faith that appeals to emo-
tion at the expense of reason. As Allen Wood explains, “Kant is will-
ing to condone a faith which bases itself on special divine revelation
only insofar as the content of its revelation accords with the precepts
revealed naturally to every human being through the faculty of rea-
son.”1 And in keeping with transcendental idealism, Kant rejects the
possibility of metaphysical knowledge of God. As he famously puts it
in the 1787 Preface: “Thus I had to deny knowledge in order to make
room for faith” (Bxxx). On the other hand, although rational theol-
ogy is a pseudoscience, the idea of God serves two legitimate purposes.
First, it is necessary for moral faith. As rational moral agents, we recog-
nize the moral law to pursue the highest good. But we can realize our
purposes only within the world of nature. Thus moral action makes
sense only on the assumption that nature is in harmony with morality.
For Kant, this implies that nature is governed by a supremely perfect
being. Kant elaborates on this point in the Lectures on the Philosophical
Doctrine of Religion as well as his ethical writings. In its second role,
the idea of God has a regulative function promoting the inquiry into
natural purposive systems in empirical science. The Critique of the
Power of Judgment contains the detailed explanation of this role. Here
his main purpose is to critique the assumptions of rational theology.
The ¬rst part of the chapter presents a rather dense account of the ori-
gin of the idea of God. In the remainder Kant makes his penetrating
analyses of the three traditional proofs of the existence of God.

1 Wood, Kant™s Rational Theology, 16.

Transcendental illusion III 265

1. th e idea l of pure re a s o n
In sections 1“3 Kant explains how the idea of God arises as an ideal of
reason. His account distinguishes two transcendental ideas: ¬rst, the
idea of the sum total of all reality, as all possible predicates of things,
and second, the idea of the ens realissimum, the individual having
the highest degree of reality. Kant calls the latter the ideal of reason.
Both ideas represent the unconditioned, in this case that underlying
all objects in general.
In section 1 Kant compares these ideas to Plato™s Forms. The idea
of “Humanity in its entire perfection” (A568/B596), for example,
is the idea of the properties essential to human nature as well as
contingent properties consistent with this idea. Like Plato™s Form of
humanity, this idea is a perfect exemplar of its type and the ground of
all (imperfect) copies in appearance. Now the idea of the individual
embodying all these perfections would be the idea of a divine human
being, such as the sage of the Stoics. Because no appearance satis¬es
either the idea or the ideal of reason, neither has objective reality.
Nonetheless, like Plato™s Forms, they have regulative signi¬cance as
standards of action and evaluation.
Section 2 explains how these ideas arise in the logical processes
involved in thinking determinate objects. Kant discusses two princi-
ples of determination, one concerning concepts, and the other existing
things. All concepts are subject to the Principle of Determinability
(PD): to determine the content of a concept is to apply one of a
pair of opposing predicates to it. This procedure is governed by the
principle of contradiction, according to which at most one of two
opposed predicates can be contained in a concept. The logical princi-
ple makes consistency a necessary condition for the form of concepts.
The second principle, that of “thoroughgoing determination” (Prin-
ciple of Thoroughgoing Determinability or PTD), applies to existing
things. This is the traditional view that every existing thing is com-
pletely determined with respect to “every pair of possible predicates”
(A573/B601). More formally, for every possible existent and for every
pair of possible predicates, one (and only one) predicate must apply to
the thing. This principle underlies the idea of the complete cognition
of a thing. But since a complete cognition is not attainable, the PTD
can never be exhibited in concreto, and thus is a transcendental idea
Transcendental illusion III
of reason. Rather than representing an object, it actually represents a
procedure for cognizing an object.
This procedure can be carried out only against the backdrop of
“the idea of an All of reality (omnitudo realitatis)” (A575“6/B604). To
af¬rm a predicate of something requires conceiving the predicate as
a kind of reality. Conceiving the absence of a reality logically presup-
poses the positive concept of the reality. Thus the idea of the sum
total of possible predicates constitutes “a transcendental substratum”
grounding all concepts of existing things. From here it is a short step
to the transcendental ideal of an individual having the highest reality.
This occurs by thinking the collective unity of all possible realities
as an individual. All concepts of individuals presuppose this ideal “
the ens realissimum “ as the ground of “thoroughgoing determination
that is necessarily encountered in everything existing.” Kant explains
this process in terms of the disjunctive syllogism, in which reason
presupposes only the idea of the being answering to the ideal, not its
Under the in¬‚uence of transcendental illusion, reason hypostatizes
the ens realissimum as an actual being having all possible reality, the ens
originarium, ens summum, ens entium (original being, highest being,
being of all beings) (A579/B607). When personalized “as a being that
is singular, simple, all-suf¬cient, eternal,” a divine intelligence and
will, this becomes the theological idea of God. Like the ideas of the
world as a whole and the soul, however, the idea of God oversteps
all bounds of experience, and thus does not represent an object of
knowledge. Kant says, “we dialectically transform the distributive
unity of the use of the understanding in experience, into the collective
unity of a whole of experience” (A583/B661). In other words, the
legitimate thought of the totality of predicates distributed among
possible objects of experience becomes the idea of the collection of
properties to be predicated of a single individual.
In section 3 Kant explains how reason then hypostatizes this ideal
by means of the transcendental illusion underlying the Paralogisms
and the Antinomies. In seeking the unconditioned, reason applies the
2 In the New Elucidation (1755) and The Only Possible Argument (1763) Kant made this argument
for the existence of God (Theoretical Philosophy, 1755“1770, 1“45 and 107“201). Wood calls it
the “possibility proof” and discusses both its pre-critical and critical uses at Kant™s Rational
Theology, 64“71.
Transcendental illusion III 267
illusory principle P2 , “If the conditioned is given, the entire series of
conditions is given,” to objects in general. Here reason searches for
an absolutely necessary being underlying all contingency: “For the
contingent exists only under the condition of something else as its
cause . . . necessarily without condition” (A584/B612). As opposed
to the Fourth Antinomy cosmological idea of a necessary being in
appearances, the theological idea represents a necessary thing in itself
underlying objects in general. Because this latter idea represents only
something whose non-being is impossible (A592/B620), it is inde-
terminate with respect to perfection, and is equally applicable to a
limited being. Nonetheless, reason naturally takes the ens realissimum
as the best candidate for an absolutely necessary being since “it satis-
¬es the concept of unconditioned necessity on at least one point . . .
since every other concept is defective and in need of completion”
(A585“6/B613“14). A reinforcing motive resides in the demands of
practical reason, since the existence of a highest being would provide
a subjective basis for obeying the moral law. In this way the natural
demand of reason for closure in the series of conditions leads humans
to argue for the necessary existence of God as the ens realissimum.
At A590“1/B618“9 Kant classi¬es the three traditional proofs for the
existence of God in terms of their evidence. The physico-theological
proof, better known as the argument from design, is based on obser-
vations of “the special constitution of our world,” and argues that
God must exist as the author of the order experienced in nature. The
cosmological argument that God exists as the creator of the world
is also empirically based, but on an “indeterminate” experience of
existence. The ontological proof differs in inferring “the existence of
a highest cause entirely a priori from mere concepts.” Because Kant
believes the two empirical arguments covertly presuppose the onto-
logical proof, he begins his criticism with that argument. In all three
cases, he argues that the proofs fail to demonstrate that God exists as
an absolutely necessary being.

2. th e ontolog i ca l a rgu me nt
Oddly enough, Kant™s discussion lacks a detailed account of the onto-
logical argument, beginning abruptly with his criticisms of it. (He
only brie¬‚y sketches the other two proofs.) So it may be helpful
Transcendental illusion III
to present the most famous versions. The argument was originally
formulated by St. Anselm (1033“1109), Archbishop of Canterbury.
Anselm bases the existence of God on the idea of God as that than
which nothing greater can be conceived. The argument as it appears
in the Proslogion is this:
For, it is possible to conceive of a being which cannot be conceived not to
exist, and this is greater than one which can be conceived not to exist. Hence,
if that, than which nothing greater can be conceived, can be conceived not
to exist, it is not that, than which nothing greater can be conceived. But this
is an irreconcilable contradiction. There is, then, so truly a being than which


. 7
( 9)