Kant's First Paralogism: Ian Proops

Download as pdf or txt
Download as pdf or txt
You are on page 1of 48

Kants First Paralogism

Ian Proops
University of Texas at Austin
Introduction
Kant pursues a variety of projects in the Dialectic of the Critique of Pure
Reason; but chief among them is the program of transcendental criti-
cism: pure reasons systematic detection, correction, and explanation
of its own excesses.
1
These excesses occur, according to Kant, because
reason constantly strives to transcend the bounds of sense in its pursuit
of speculative knowledge. In Kants view, all such attempts to rise above
the world of sense throughthe mere might of speculation (A591/B619)
are destined to fail.
2
In particular, pure speculative reason can afford us
My thanks to Omri Boehm, Jim Hankinson, Ivan Heyman, Martin Lin, Dustin Locke,
Christopher Peacocke, Michael Della Rocca, Anat Schechtman, Nicole Smith, David
Sosa, and Zolt an Szab o. Special thanks are due to Kenneth Winkler, who read several
drafts of this essay and provided astute and generous comments on each draft. One
ancestor of this essay was presented at the University of Texas in November 2007; another
was discussed at a lunchtime meeting of the philosophy faculty at Yale University in the
spring of 2008. I amgrateful to those present on each occasion for their many helpful and
thought-provoking questions. My thanks also to two anonymous referees for this journal
who provided indispensible comments whose infuence appears on almost every page.
1. Kants other goals include the promulgation of his doctrine of the regulative use
of the Ideas of Reason (A 508/B 536ff.) and the rehearsal of his indirect argument (in
the Antinomies chapter) for Transcendental Idealism (A 490/B 518ff.).
2. For passages from the frst Critique, I have usually followed the translation by Paul
Guyer and Allen Wood (1998). The points on which I have offered my own translation,
when it concerns a substantive rather than merely stylistic alteration, are indicated by
interpolating the relevant German word or phrase and by adding translation my own
in parentheses. The basis for these translations is Kant 1902. Passages from this work are
cited in the text by the relevant volume and page number. When I mark a translation
as my own, it often departs only in minor respects from the Guyer-Wood translation,
Philosophical Review, Vol. 119, No. 4, 2010
DOI 10.1215/00318108-2010-011
2010 by Cornell University
449
I A N P R O O P S
no knowledge of the nature of the self as it is in itself, none of the world
considered as the totality of appearances, and none of the existence (or
nonexistence
3
) of God. Instead, the content of our speculative knowl-
edge is confned to those principles that must be true if experience is to
be possible, and such principlesKant supposesspeak only of objects
of possible experience.
4
Since the views Kant criticizes as dogmatic in the Dialectic had
been propounded by some of the brightest minds of the Early Mod-
ern period, he owes his reader a suitably charitable explanation of how
thinkers of the caliber of, for example, Descartes and Leibniz could have
gone as far astray as he contends.
5
Recognizing this obligation, Kant elab-
orates a novel and ingenious account of the seductiveness of dogmatic
error. The rationalist philosophers go wrong, he maintains, not because
they fall prey to the artful invention of sophistsand certainly not
because of any blundering intellectual clumsiness on their part (A 298/B
354)but simply because they succumb to a powerful illusion grounded
in the very nature of human reason itself (A 298/B 354).
6
Because he
takes even the most acute of human minds to be subject to this illu-
sion,
7
Kant can claim not to be implausibly charging his opponents
and I do not mean, in those instances, to imply a lack of indebtedness to them for the
remainder of my translation.
3. Kant believes that we will not be able to prove the nonexistence of God, say, by
discovering a contradiction in the concept of God. This feature of his view is an aspect
of his so-called doctrine of practical faith. If we are to be moral, we must believe God to
exist and, a fortiori, to be possible. This point is clear from a refection from 178384:
In moral theology it is enough to presuppose that it is still possible that there is a God,
and that no one can ever prove the non-being thereof . R 6236; 18:51920, emphasis added.
The surrounding context makes clear that by moral theology Kant means the body of
beliefs that we must hold if we are to be motivated by the moral law.
4. See B 294; A 471/B 499; 11:344, and compare B 289; A 7023/B 73031; 4:336n.
5. Kants immediate targets in the Paralogisms include both Leibniz himself and his
followersWolff and Baumgarten in particular, and as Kant himself makes clear at B 413,
on the issue of the immortality of the soul: Mendelssohn. Leibniz, as Margaret Wilson
has argued, seems to be Kants target specifcally in the second paralogism. Descartes,
for his part, is explicitly identifed as a target in the fourth paralogism. For details, see
Longuenesse 2008, 20; and Wilson 1974.
6. In the case of Leibniz, this account of the source of metaphysical error is supple-
mented by a separate account, offered in the Amphiboly to the Concepts of Refection,
of how the empirical use of the understanding can be confused with one that seeks to go
beyond experience. A 260/B 316A 292/B 349.
7. The illusion, Kant says, can fool even the most rational. A 703/B 731.
450
Kants First Paralogism
with navet e or sloppiness. Kant calls this supposedly pervasive illusion
transcendental illusion.
In Kants view, this distinctively intellectual species of illusion is
not by itself suffcient to account for the existence of dogmatic meta-
physics. It explains the pervasive and abiding temptation to dogmatism,
but the actual commission of dogmatic error is explained only with
the help of further resources. We form dogmatic views, Kant supposes,
when, in the grip of transcendental illusion, we allow ourselves to be
persuaded of the soundness of certain unsound arguments, which Kant
terms dialectical inferences of pure reason (A 338/B 396). These argu-
ments divide into three groups: The paralogisms are unsound argu-
ments for various dogmatic theses about the nature of the self. The
antinomiesor at least the mathematical subspecies of them
8
are
pairs of unsound arguments purporting to establish opposed doctrines
about the world considered as the totality of appearances. Finally, the
arguments of the Ideal of Pure Reason are unsound arguments that
lead us dogmatically to assert the existence of God, conceived abstractly
as the ens realissimuma being that is supremely perfect in the sense
of possessing the maximum possible degree of reality.
This account of Kants diagnosis of dogmatic error acknowledges
the importance of a distinctionoriginally drawn by Michelle Grier
9

between the phenomenon of transcendental illusion, on the one hand,


and the unsound arguments to which it disposes fnite rational beings,
on the other. The need to draw such a distinction is, to my knowledge, by
now broadly accepted by Kant scholars. More controversial, and less well
understood, are some of the details of how this diagnostic model applies
to the specifc doctrines criticized in the Dialectic. In particular, there
is little agreement about how this model applies to Kants criticisms of
various rationalist doctrines about the self in the Paralogisms chapter.
The present essay seeks to solidify our understanding in this area
by applying Kants diagnostic model to the case of the self. Or, rather,
it narrows its focus still further and seeks to apply this model to just
one aspect of the Cartesian or Leibnizian dogmatists theory of the self,
namely, the view that the self is a substance. This task, I submit, is worth
8. In Kants view, the third and fourth antinomies, which are known as the dynam-
ical antinomies, do not comprise pairs of unsound arguments, for they are resolved by
arguing that thesis and antithesis, taken the right way, can both be true. The chief point is
that in these cases the thesis and antithesis only seem to be logically inconsistent because
of the transcendentally realist assumptions encouraged by transcendental illusion.
9. I discuss Griers work in notes 11 and 51 below. Citations are provided in note 19.
451
I A N P R O O P S
undertaking both for the light it sheds on Kants conception of meta-
physics as practiced by his immediate predecessors and for what it reveals
about his conception of philosophical criticism. It should also be of inter-
est for what it reveals about Kants take on certain venerable philosophi-
cal notions, including the very notion of substance. We shall see that, in
order to do justice to the richness of Kants discussion, it will be necessary
to distinguish no fewer than four rather different Kantian conceptions of
substance.
10
Our discussion recapitulates the bifurcation in Kants diagnosis
of dogmatism: its frst part provides an account of how transcendental
illusion creates a temptation to believe that the self is a substance, while
its second offers an analysis of the fallacy that Kant supposes we com-
mit when we succumb to this temptation. At higher resolution, the essay
divides into seven sections and a conclusion. The frst section discusses
the phenomenon of transcendental illusion in general; the second asks
how this illusion is supposed to generate in human beings an inclina-
tion to believe in the substantiality of the soul. The third asks what pre-
cisely Kant means by a paralogism, while the fourth examines the B-
editions diagnosis of the formal fallacy involved in the frst paralogism.
The ffth section examines the rather different account of the frst paral-
ogism given in the A-edition and offers an account of why Kant came
to see his earlier discussion as unsatisfactory. The penultimate section is
devoted to defending my interpretation of the B-paralogism against pos-
sible objections, and the last to examining some further complications
of the story told in the earlier sections. Along the way, I will identify and
discuss a number of as yet unrecognized diffculties for Kants account
of the frst paralogism and for his account of transcendental illusion in
general. In spite of these problems, I hope to show that Kants views on
these matters are more promisingand, indeed, more philosophically
illuminatingthan many commentators have supposed. I will argue, in
particular, that Kants discussion of the B-editions frst paralogism yields
an insight about the nature of conceivability that is of lasting value.
1. Transcendental Illusion
As Kant explains it, transcendental illusion is the illusion that what are
in fact merely so many maxims or rules guiding empirical inquiry amount
10. The frst three conceptions are discussed in the mainbody of the essay. The fourth
is discussed in note 54.
452
Kants First Paralogism
to objective truths (compare A 297/B 353). These rules, which human
reason provides for its own direction, are instances of a more general
demand of reason, which Kant formulates thus:
P: Find for the conditioned cognitions of the understanding the
unconditioned whereby its [that is, the understandings] unity
is completed (compare A 307/B 364).
Such a prescription enjoins us to seek, among other things, ever more
fundamental explanations and ever earlier causes for the phenomena
we encounter in experience.
Kant maintains that, owing to transcendental illusion, rational
human beings are inclined to take this logical prescription (A 309/
B 365) for the descriptive claim that:
D: If the conditioned is given, the whole series of conditions sub-
ordinated to one another, a series which is therefore itself
unconditioned, is likewise given (compare A 3078/B 364 and
B 436).
11
When through a misunderstanding we succumb to transcendental illu-
sion, the need of reason expressed by P is taken for a transcendental
principle of reason that over hastily postulates . . . an unlimited com-
pleteness in the series of conditions (compare A 309/B 366).
12
In other
words, the prescription, P, is taken for the putative statement of fact,
11. In seeing transcendental illusion as consistingas a frst approximationin the
illusion that D is true, I amin general agreement with Griers illuminating account of this
phenomenon in Grier 1993 and 2001. (My labels P [for prescription] and D [for
description] are merely intended as perhaps more suggestive replacements for Griers
P1 and P2.) Although I am in broad agreement with much of what Grier saysand
in particular on the general need to distinguish transcendental illusion from error, I dis-
agree with her on certain points of detail. First, I would not endorse her characterization
of D as something we must assume as a condition of using P. Grier 2001, 126. For,
being a prescription, P is not something that can be used but only something that can be
complied with or violated. On my account, the correct thing to say about the relationship
between P and D is rather that the illusion that D is true is one that is inevitably generated
in fnite rational minds in virtue of their being subject to the demand of reason expressed
by the prescription P. Second, as we shall see in what follows, I believe that in the Paral-
ogisms it is not quite D itself but a closely related claim that expresses the content of the
illusion (see section 7). Third, I differ from Grier in my interpretation of the argument
form of the frst paralogism (see section 6 below, and especially note 51).
12. Kant gives this explanation in the form of a question in the course of announcing
the topic to be investigated. The question is lengthy and hard to parse, but we might
paraphrase it thus: Has the need of reason expressed by P been taken for the transcen-
dental principle expressed by D? It is clear from his subsequent discussion that, as far as
traditional rationalist metaphysics goes, Kants answer is yes.
453
I A N P R O O P S
D. In Kants jargon, a regulative principle is taken for a constitutive
principle.
13
Although Kant characterizes the misunderstanding (A 309/B
365) that leads to the fallacious inferences of reason in terms of our
having taken the need of reason expressed by the prescription P for the
truth of the principle D (A 309/B 366),
14
the path fromthe former to the
latter can be fruitfully described in more subtle terms. It seems plausible
that, being a command of reason, P must be viewed as an authoritative
prescription. It is also plausibly experienced as a reasonable prescription
since reasons authority strikes those subject to it as reasonably exercised.
But since the prescription, P, strikes us as reasonable and authoritative,
it is extremely natural to suppose that what it enjoins us to seek is in fact
there to be found. Thus our consciousness of the standing authority of
P generates in us the permanent inclination to regard D as true. We
may, therefore, characterize transcendental illusion simply as the illusion
that D is truean illusion that is explained by our being subject to the
prescription P.
Or so, at least, runs a plausible interpretation of Kants frst-pass
view. However, I shall argue in section 7 that Kants more considered view
is that it is in fact not D itself that states the content of transcendental
illusionbut rather the related propositionD: If the conditioned is given,
the unconditioned is likewise given. However, because the reasons for
this qualifcation are rather complicated, it will be useful, for the time
being, to continue our exposition with D.
15
The phraseology of P and D immediately raises a number of ques-
tions: Are the members of the series of conditions alluded to by P and
D to be understood as cognitions (as the phrasing of P might seem to
suggest), or are they rather to be thought of as the various states, events,
13. See A 61920/B 64748; A 647/B 675; A 686/B 713; A 690/B 718; and Prolegom-
ena, sec. 56, 4:350.
14. Kant also implies that instances of P are apt to be taken for instances of D, for
he says that the various specifc fundamental maxims and rules contributed by human
reason and governing its use look entirely like objective principles. A 297/B 313.
15. In characterizing the governing illusion behind all of rationalist metaphysics as
the illusion that D is true, Kant may seem to be lumping his predecessors together in
a rather high-handed way. Is it really true that all of rational metaphysics fows from
a single principle? Such qualms may, however, be allayed somewhat by Omri Boehms
observation (Boehm 2009) that D can be regarded as one formulation of the Principle
of Suffcient Reason, a principle that does indeed have a claimto be the single substantive
principle generating all rationalist metaphysical claims. This idea might also explain why
Kant refers to D as this supreme principle of pure reason. A 308/B 365.
454
Kants First Paralogism
or properties that are the objects of cognitions? What is it for a condition
to be given? And, perhaps most baffing of all: What would it mean for
the unconditioned to complete the unity of the understanding?
My interpretation answers the frst two of these questions as fol-
lows: First, the conditions in question are not cognitions but objec-
tive states of the world (or, in some cases, events or properties). Thus in
enjoining us to seek the unconditioned for the conditioned cognitions
of the understanding, P is enjoining us always to seek ever more funda-
mental conditions for the objects of our cognitions. Second, to speak of a
condition as being given is just to say that it exists (or obtains). I shall
not attempt to argue for these claims here as the justifcation for them is
just that they make possible a coherent and plausible interpretation both
of transcendental illusion itself and of the way this illusion plays out in the
Antinomies and the frst paralogism. The third question does, however,
merit a more detailed answer.
In order to understand the notion of completing the unity of the
understanding, it helps to note two relatively obscure but still important
details of the Leibnizian background. First, in Baumgartens Metaphys-
icathe Leibnizian work on which Kant usually based his metaphysics
lecturesthe term condition is treated merely as a variant of ground
(see Baumgarten 1902, sec. 14). Kant seems to follow Baumgarten on
this point, for he occasionally speaks indifferently of grounds or condi-
tions (see, for example, Real Progress, 20:328). The second point to note
is that, in Baumgartens view, grounds and conditions are precisely what
one cites in answering why-questions (Baumgarten 1902, sec. 14). Kant
shares this opinion (Vienna Logic, 24:921), but he takes issue with Baum-
gartens claim that one can defne a ground as that from which it can be
cognized why something is. Such a defnition, Kant contends, would be
circular because why just means through which ground (Vienna Logic,
24:921). Importantly, however, while he rejects Baumgartens attempted
defnition as circular, he still regards it as true (Vienna Logic, 24:921). So,
for Kant, whenever one states a ground or condition, one thereby answers
a why-question. Taken together, these observations suggest that Kant is
thinking of the demand that we seek the unconditioned condition for
something as the demand that we seek something which, if found, would
answer the fnal answerable why-question in a series of such questions.
If that is right, then to have found an unconditioned condition would
be to be possess the most complete understanding possible of the original
phenomenon to be explained. In Kants language, one would have com-
pleted the unity of the understanding.
455
I A N P R O O P S
In Kants view, we are inclined to regard D not just as true but
as necessarily true. This circumstance, he maintains, is owed to our readi-
ness to mistake the subjective necessity of a certain connection of
our concepts for an objective necessity, the determination of things in
themselves (A 297/B 353). Kant does not explain what this subjective
necessity consists in; but we might plausibly think of it as the necessity
involved in the following conditional injunction: If you wish to proceed
rationally in inquiry, you must seek, for the object of every cognition, the
series constituting its unconditioned condition. The necessity involved
here counts as subjective because the imperative is conditional upon
our having a certain desire. Should one not wish to proceed rationally in
inquiry, one will stand under no obligation of the kind specifed in the
conditionals consequent.
Kant supposes that, owing to transcendental illusion, this subjec-
tive necessity is apt to be misconstrued as the objective necessity invoked
by the claim: Necessarily, for the object of any cognition, the series con-
stituting its unconditioned condition exists. The illusion that D holds of
necessity is thus itself a component of transcendental illusion. The slide
Kant envisions here may be made clearer by breaking it down into two
transitions. Consider the following three statements:
[A] If you desire to proceed rationally in inquiry, you must seek,
for the object of any cognition, the series constituting its un-
conditioned condition.
[B] You must seek, for the object of any cognition, the series con-
stituting its unconditioned condition.
[C] Necessarily, for the object of any cognition, the series con-
stituting its unconditioned condition exists.
Claim [C] is just the claim that principle D (slightly reformulated) holds
of necessity. As I read him, Kant supposes we are apt to confuse [A]
with [B] simply because, as rational beings, we naturally desire to pro-
ceed rationally in inquiry. Because this desire is so natural and pervasive,
its presence is easily overlooked, and we thus fail to appreciate that the
necessity in [B] is merely the conditional necessity expressed in the con-
sequent of [A]. Once we have mistaken [A] for [B], transcendental illu-
sion inclines us to slide further to [C]: for we hear the injunction of [B]
as an injunction of reason, and so regard it as rational. This in turn leads
us to believe that what we are enjoined to seek actually exists. We thus
misconstrue the must of a binding intellectual categorical imperative
456
Kants First Paralogism
that is, [B]as the must of an indicative claim that purports to express
a necessary factthat is, [C].
In support of this explanation, we may observe that Kant himself
on one occasion draws the distinction between objective and subjective
necessity in precisely these termsalbeit in a somewhat different con-
nection. In his Lectures on the Philosophical Doctrine of Religion, which stem
from the mid-1780s, Kant remarks regarding his attempted proof of the
ens realissimum (or most real being) in his 1763 work The Only Possible
Basis for a Demonstration of the Existence of God that this proof
is unable to establish the objective necessity of an original being; rather it
establishes only the subjective necessity of accepting such a being. But this
proof can in no way be refuted, because it has its ground in the nature
of human reason. For my reason makes it absolutely necessary for me
to accept a being which is the ground of everything possible, because
otherwise I would be unable to cognize what in general the possibility of
something consists in. (28:1034)
16
Kant is warning us here not to take his 1763 proof to establish
the objective necessity of the claim that an ens realissimum exists. Rather,
it should be taken to establish the merely subjective necessity involved in
the conditional claim: If we want to cognize in what in general the possi-
bility of something consists, then we must accept that the ens realissimum
exists. Having noted this parallel, we should concede that in the case of
the hypothetical imperative governing inquiry, [A], what fgures in the
consequent of the conditional is not the notion of accepting a proposi-
tion but rather that of seeking the unconditioned. Nonetheless, this dif-
ference does not undermine the parallel between these passages insofar
as it relates to Kants use of the terms subjective and objective.
Kant conceives of transcendental illusion as sharing three fea-
tures with cases of ordinary perceptualor empirical (A 295/B 351)
illusion. First, it is natural and unavoidable (unvermeidlich) (A 298/B 354):
just as anyone who possesses a properly functioning visual faculty will,
Kant supposes, be subject to the moonillusion;
17
so anyone who possesses
a properly functioning faculty of reason will experience transcendental
illusion. Presumably because he wishes to exempt a possible divine reason
16. This passage is cited in a fascinating essay by Andrew Chignell, 2007, 349. I am
indebted to him for raising it to prominence, and I follow his translation.
17. The moon illusion refers to the illusion that the moon is larger on the horizon
than it is at its apex. Kant assumes that such optical illusions are universal (A 297/B 354);
whether that is in fact so is, of course, a question for empirical psychology.
457
I A N P R O O P S
from transcendental illusion, Kant tends to characterize this kind of illu-
sion as inevitable for human reason. Such an exemption would be well
judged since a divine rational faculty, should it exist, would need to be
conceived of as immune fromtranscendental illusion on pain of compro-
mising the Divinitys perfection. (Sometimes in what follows the qualif-
cation human will be omitted, but only for stylistic reasons.)
The second feature transcendental illusion shares with perceptual
illusion is that its usual consequences are correctible:
18
just as one need
not be taken in by an optical illusion in the sense of actually forming
an erroneous judgment or belief on its basis, so one need not be led
into actual error by transcendental illusion. As Grier has emphasized,
19
this distinction between unavoidable illusion, on the one hand, and
avoidable error, on the other, is essential for understanding Kants criti-
cal enterprise: without it Kant would seem to be making the manifestly
implausible claimthat the unsound arguments themselves are inevitable.
Thus, rather than urging us to the impossible feat of avoiding inevitable
sophistry, Kant is aiming to help us to avoid unsound arguments, even
while insisting that the illusion that inclines us toward fnding them com-
pelling cannot be removed.
This brings us to a third feature that transcendental illusion shares
with empirical illusion, namely: its persistence. Transcendental illusion
does not cease even though it is uncovered and its nullity clearly seen
into by transcendental criticism (A 297/B 353). In other words, tran-
scendental illusion survives detection and even explanation. Owing to
transcendental illusions persistence, human reason is continually pro-
pelled into momentary aberrations that always need to be removed
(A 298/B 355). Accordingly, the work of transcendental criticismis never
complete.
Kants account of transcendental illusion is clear; but how does it
play out in the case of a particular transcendental idea? This is easy to
see in the case of the transcendental idea of the world, which is associated
with certain argumentsthe antinomiesthat purport to establish con-
clusions about the world as a whole. In the frst antinomy, for example,
18. I avoid saying that the illusion itself is correctible since this might create the mis-
leading impression that the illusion is capable of being neutralized by a counterillusion,
in the way that, for instance, some forms of corrective lens provide a compensating
counterillusion. I also avoid saying that the illusion is resistible since we cannot get rid
of ita point, indeed, that Kant puts by saying that it is irresistible [unwiderstehlichen].
A 642/B 670.
19. See Grier 1993, 26364 and Grier 2001, 116 and 128ff.
458
Kants First Paralogism
transcendental illusion tempts us to embrace both the belief that the
world must have had a beginning (the thesis
20
) and also the apparently
contradictory belief that the world cannot have had a beginning (the
antithesis).
21
And, indeed, it tempts us to view these beliefs as amounting
to rationally grounded knowledge. In the case of the thesis, the whole
regressive series of conditionsin this context, ever earlier states of
the worldis taken to terminate in an earliest condition of the series,
which is itself unconditioned (compare A 418/B 446). In the case of the
antithesis, on the other hand, the regress is taken not to terminate, but
the whole infnite series without its frst membernamely, the present
state of the worldis taken to be the unconditioned condition of that
frst member (compare A 418/B 446, including footnote

). The frst
antinomy arises because we thus haveor, rather, imagine we havetwo
incompatible ways of fnding the completeness that reason enjoins us to
seek in the series of conditions. Dogmatic theorizing about the age (and
size) of the world will thus lead to unending controversies since its prac-
titioners will divide between these apparently incompatible alternatives
according to their individual proclivitiesincluding their taste, or lack
of taste, for theism.
22
20. Or, strictly speaking, part of it, since the antinomy also concerns the question
whether the world is bounded in space.
21. Kants solution to the frst two antinomies consists in arguing that thesis and
antithesis are contraries rather than contradictories and that both are false.
22. Kants account of what I have called these proclivities is in fact rather involved,
and it has an interesting philosophical motivation. He supposes that if we could consider
the thesis and antithesis positions purely on their intrinsic meritsthat is, according
to their grounds (A 475/B 503), we would be led to an unceasing vacillation. Ibid.
In fact, Kant supposes that we would fip-fop between thesis and antithesis on a daily
basis. Ibid. In practice, however, individuals who consider the antinomial questions tend
to form themselves into relatively stable camps or factions. In one corner, favoring the
thesis position, are the dogmatists (compare A 466/B 494) or advocates of Platonism
(A 471/B 499). In the other, favoring the antithesis, are the empiricists (A 468/B 496)
or champions of Epicureanism (A 471/B 499). In a broader sense, of course, both
camps are dogmatic, and it is clear that Kant views his labels for these positions as loose
and possibly historically inaccurate. A 471/B 499.
Although Kant doesnt himself explicitly mention the point, this apparent mismatch
between Kants theory of transcendental illusion, which predicts extreme doctrinal insta-
bility, and observation, which reveals a measure of stability, constitutes a prima facie dif-
fculty for his theory. To address it, Kant offers a supplementary explanation: he suggests
that in practice certain factors unrelated to the intrinsic merits of the claims infuence
our choice between them. These factors, which Kant terms interests of reason, are as
follows: First, there is a certain practical interest of reason, namely, the need to employ
the four thesis positions as cornerstones of religion and morality. A 466/B 494. Second,
459
I A N P R O O P S
This is only the briefest sketch of a good deal of complex mate-
rial, but it does, I would claim, capture the general contours of Kants
application of the model of transcendental illusion to the case of the
Antinomies. Unfortunately, it takes considerably more effort to see how
this model might apply to the case of the self as it is discussed in the
Paralogisms. It is to this question that we now turn.
2. Transcendental Illusion and the Self
The proponent of rational psychologythe rationalists putative sci-
ence of the selfpresumes that we can know the character of the self
as a simple, incorruptible, immaterial, and naturally immortal substance
(A 345/B 403, A 349)a soul as conceived by the Christian tradition.
This traditional conception of the soul should not be confused with
Kants ownmore abstract or formal conception, according to whicha soul
(Seele) is just a thinking being (ein denkend Wesen) (A 348)where the
notion of a being is maximally thin and does not, for example, entail
substantiality.
23
Kant maintains that transcendental illusion explains the
prevalence of the belief in the richer, traditional conception of the soul
because it unceasingly tempts us to draw certain fallacious inferences
about the character of this thinking being. In Kants view, however, the
there is the interest of popularity, which favors the thesis becauseKant supposesthe
common understanding has no scruples about the idea of an unconditioned. A 467/B
495. One supposes that Kant would viewthis lack of scruples as owed, at least inpart, to the
ordinary persons ignorance of the Principle of Suffcient Reason. Third, there is a certain
architectonic interest,a kind of penchant for system building and foundationalism
which favors the thesis of each antinomy. A 47475/B 503. A fourth interest of reason
does not so much stabilize the oscillation as fuel it. This is the speculative interest of
reason, which can work in favor of either the thesis or the antithesis. A 46667/B 49495.
Kant is not explicit about the nature of this interest, but he may well have in mind the
demand of reason expressed by P, which, as we have seen, can be satisfed in either of two
incompatible ways. (I am grateful to an anonymous referee for this suggestion.) Notice
that since each of the stabilizing interests is supposed to favor the thesis position, Kant is
still left with a residual problem of explaining the stability of the antithesis position.
23. Kant uses a range of expressions for his own, much thinner, notion of the soul
in the frst Critique. These include: the soul [die Seele] (A 351; A 361); the thinking I
[das denkende Ich] (A 351; A 361; A 401); the thinking self [das denkende Selbst] (B 430),
and [the] transcendental subject of thoughts. A 346/B 404. At A 361 he suggests that
the frst of these expressions refers to the transcendental object of inner sense, which
presumably means the self that appears to itself in inner sense as the empirical self.
Kants remarks at B 429 are consistent with taking the thinking self at B 430 to have the
same reference. Thus the thinking I or thinking self is both the transcendental subject
of thoughts and the transcendental object of inner sense.
460
Kants First Paralogism
thinking being that we refer to by means of the word I is something
about whose inner nature we must remain forever ignorant. (In what
follows I will use the terms the self and the soul indifferently when
reporting Kants views about this thinking being.)
In the case of the B-editions frst paralogism, transcendental illu-
sion is supposed to encourage the belief that the self is a substance in the
sense of something that can exist only as subject, not as a mere determi-
nation of other things. (B 288; compare B 149, B 289). A substance in
this sense is something that cannot inhere in another thing as a deter-
mination of it. And by a determination in this context Kant means
something that is either a property or a mode. (The latter is an entity
whose existence consists in the exemplifcation of a property or relation
by another thing [or other things]. Common examples include dents
and fsts.
24
) Kant also emphasizes that it is something that need not so
inhere because it is a self-subsisting being (B 407; compare B 412, B
149). In the B-edition, Kant intends to be speaking of substance in this
sense whenever he invokes the unschematized category of substance;
but in both editions he also appeals to a distinct concept of substance
corresponding to the schematized category. This is the concept of
something that persists (A 144/B 183, A 182) in the sense of existing at
every moment (A 182). To keep ideas straight, I will follow Jonathan Ben-
netts convention of referring to the frst of these notions as substance
1

and to the second as substance


2
(Bennett 1966, 182ff.). Even though
Kant uses the same term, Substanz, for both notions, he is good at not
confusing them.
Kant is clear that we cannot know whether or not there are any
substances
1
or even whether or not their existence is a real possibility:
[In the General Note on the System of Principles (B 288B 294) and
the chapter on Phenomena and Noumena (A 23560/B 294B 315)] it
was proved that the concept of a thing that can exist for itself as subject
but not as mere predicate [als bloes Pr adikat] carries with it no objective
reality at all, that is, that one cannot know whether it can apply [zukommen
k onne] to any object, since one has no insight into the possibility of such
a way of existing, and consequently that it yields absolutely no cognition.
(B 412, translation my own).
Because the concept of a substance
1
contains a mark of necessity (namely,
cannot occur as a predicate), which no experience can yield, we are
24. For this understanding of the notions of determination and mode I am
indebted to James van Cleves discussion in van Cleve 1999, 105.
461
I A N P R O O P S
incapable of having an intuition of a substance
1
, and so cannot thereby
know the concept of substance
1
to be capable of having instancesthat
is, to be objectively real. Nor can we know it to have instances in the
way that, according to Kant, we can (and do) know that (say) the con-
cept of cause has instances; for being an unschematized category, the con-
cept of substance
1
will fgure in no a priori synthetic principle of expe-
rience that could guarantee its actuality. For these reasons, we cannot
know there is not some unforeseen real repugnance between the deter-
minations represented by the various component marks of the concept
of substance
1
, and thus we cannot know whether or not a substance
1
is
a real possibility. Nor, as matters stand, do we have any reason to sup-
pose that substance
1
does not exist, for its concept seems (as far as we
can tell) to be logically consistent. To avoid dogmatism, then, we must be
agnostic about the existence or nonexistence of any substance
1
and yet
transcendental illusion tempts us to assert, dogmatically, that there is at
least one.
25
In order to apply Kants model of transcendental illusion to the
case of substance
1
, we shall need to arrive at a plausible specifcation of
the series of conditions that reason is, in this case, supposed to enjoin us
to complete. We saw that, in the case of the worlds existence in time, the
relevant series is composed of temporally ordered states of the world. In
the present case, a very different series must be under consideration. But
of what kind of things would it be composed?
In the frst Critique Kants view on this matter is hard to discern,
but in the Prolegomena it is spelled out clearly:
Pure reason demands that for each predicate of a thing we should seek
its own [zugeh origes] subject, but that for this subject, which is in turn
necessarily only a predicate, we should seek its subject again, and so
forth to infnity (or as far as we get). But from this it follows that we
should take nothing that we can attain for a fnal [letztes] subject, and
that the substantial itself could never be thought
26
by our ever-so-deeply
25. I owe a debt of gratitude to an anonymous referee for much of the substance
of this paragraph. James van Cleve (1999, 173) attributes to Kant the view that the A-
edition frst paralogism succeeds in establishing that the self is a substance
1
. However, as
will shortly become apparent, this is to attribute to Kant a position he would have deemed
unacceptably dogmatic in both editions of the frst Critique.
26. Here one expects cognized or known rather thanthought. Fortunately, how-
ever, a case can be made that Kant sometimes uses thought where he means cognized
or known. Consider, for example, a remark from the amphiboly of the concepts of
refection: If by merely intelligible objects we understand those things that are thought
462
Kants First Paralogism
penetrating understanding, even if the whole of nature were laid bare
before it. (Prolegomena, sec. 46; 4:333, translation my own)
The relation of predicate to subject generates a series in which each sub-
ject can be conceived of as (really) a predicate of a further subject that
conditions or grounds it. Kant, moreover, is clear that the regressive series
of conditions in question here is supposed to stand on all fours with the
various regressive series comprehended under the cosmological idea
(that is, the idea of the world). He makes this point in passing when, as it
happens, denying that the theological idea generates a similar series:
[In the case of the theological idea] reason does not, as with the psycho-
logical and the cosmological idea, start from experience and become seduced by the
ascending sequence of grounds into aspiring, if possible, to absolute completeness in
their series, but instead breaks off entirely from experience and descends
from bare concepts of what would constitute the absolute completeness
of a thing in general . . . to the determination of the possibility, hence the
reality, of all other things. (Prolegomena, sec. 55, 4:348, emphasis added)
The passage quoted from Prolegomena, sec. 46 makes clear what kind of
regressive or ascending series Kant has in mind in connection with the
self: it is a series of ever more fundamental subjects (that is, bearers of
properties).
27
However, while helpful, the passage is also potentially mis-
leading. It can create the impression that Kant means to be discussing
a demand of reason that might be expressed in the injunction: For
each subject, seek another in which the frst inheres as a predicate. But
although such a picture is certainly suggested by Kants words, it can-
not be quite what he intends. For this injunction implies that we should
never rest with the idea that weve found an ultimate subject. In conse-
quence, such an injunction would tend to generate in minds subject to
transcendental illusion the belief that there is no ultimate subject, and
through pure categories, without any schema of sensibility, then things of this sort are
impossible. A 286/B 342. If we took this remark at face value, it would sit awkwardly
with Kants view that the concept of a noumenon, taken merely problematically, is an
unavoidable concept that limits sensibility. A 256/B 311. For that seems to imply that
noumena are at least conceivable. Interestingly, however, Kant altered his own copy of
the frst edition to read not are thought but are cognized by us (see A 286/B 342,
Erdmann 1881, remark 150, 46; 23:49); but, having done so, he then neglected to incor-
porate this correction into the second edition. Since that is so, we may suppose that
occasionally when he uses thought in a published work, he means cognized.
27. Here we need to keep in mind that accidentswhich Kant usually refers to simply
as determinationscan be considered subjects relative to further, less fundamental,
accidents that inhere in them.
463
I A N P R O O P S
hence that nothingnot even the selfis a substance
1
. But such a belief
is obviously not Kants target in the frst paralogism; for he means to be
criticizing and diagnosing our belief that there is at least one substance
1
,
namely, the self.
A better story, which still preserves the intended parallel with the
cosmological idea, might run as follows. For each subject we encounter,
we are enjoined to seek some ultimate subject in which the associated
regressive series of subjects might be thought to terminate. Accordingly,
we must try to construe each subject we encounter as in turn a predicate
(that is, property or mode) of something else, hoping as we do so that
at some stage our attempt will be unsuccessful precisely because we will
have arrived at an ultimate subject. In obeying the injunction to proceed
in this manner we will of course be acting as if we are trying to show that
there are no ultimate subjects; but we will be doing so only as a means to
discovering a subject that resists our efforts to construe it as a predicate
of something else. Our ultimate goal, then, will be to discover an ultimate
or absolute subject. Understanding the injunction in this way yields the
satisfactory result that the dogmatic belief generated when we are taken
in by the relevant instance of transcendental illusion will be the belief
in the existence of some ultimate subjecta substance
1
in Bennetts
jargon.
Now, although Kant is certainly guilty of some carelessness here, I
think he does, nonetheless, appreciate that this is how his account ought
to run. For the discussioninthe Prolegomena continues ina way that agrees
with this account:
Now it does appear as if we have something substantial in the conscious-
ness of ourselves (i.e., in the thinking subject), and indeed have it in an
immediate intuition; for all the predicates of inner sense are referred to
the I as subject, and this cannot again be thought as the predicate of some
other subject. It therefore appears that in this case completeness in the
referring of the given concepts to a subject as predicates is not a mere
idea, but that the object, namely, the absolute subject itself, is given in
experience. But this expectation is disappointed. (Prolegomena, sec. 46,
4:334, translation my own)
Here Kant is suggesting that the injunction to seek an ultimate subject
generates the belief that there is such a thing in the case of at least one
series of subjects. In the case of the series apparently terminating in the
self , we seem to have a subject that cannot be conceived of as a pred-
icate of another thing. But this appearance, he adds, gives rise to an
464
Kants First Paralogism
expectation that is disappointed: we expect to encounter a substance
1
in introspection, but no intuition of such a thing is forthcoming. Kant
is thus concluding this remark with a Humean point about the failure
of introspection to reveal a self or soul with the propertiesof simplicity,
endurance, substantiality, and so forththat are commonly attributed to
it by rationalist metaphysicians.
The theory of transcendental illusion predicts that, other things
being equal, each of us (insofar as we are rational) will be inclined to
believe that there are substances
1
; but Kant supposes that we shall actu-
ally succumb to this inclination only when a plausible yet fallacious argu-
ment accompanies this illusion. In the case of our belief that the self is
a substance
1
, this argument is the frst paralogism. But what exactly is a
paralogism? And why does Kant treat the paralogisms as a distinctive
class of dialectical inferences?
3. The Notion of a Paralogism
It is a simple matter to describe what the arguments grouped under the
heading of The Paralogisms of Pure Reason are supposed to have in
common: they are, one and all, supposed to be sophistical categorical
syllogisms (A 4056/B 43233), whose status as sophistical stems from
their being false due to their form (A 341/B 399). Each, moreover,
is supposed to be a transcendental rather than a logical paralogism,
where this means that there is a transcendental ground for inferring
falsely due to [the arguments] form (A 341/B 399). This transcenden-
tal ground is, presumably, transcendental illusion.
28
It is, however, harder to say what exactly Kant means by the term
paralogismand harder still to answer this question in a way that
respects his idea that the Paralogisms constitute a distinctive class of
28. Although Kant unquestionably maintains that the dialectical inferences are
grounded in transcendental illusion, he also sometimes says that they ground the illusion.
For example, in the very frst sentence of the Antinomies chapter, Kant announces that
his goal is to show that every transcendental illusion of pure reason rests on dialectical
inferences [auf dialektischen Schl ussen beruhe]. A 405/B 432, emphasis added; compare A
606/B 634. To take account of this remark, one needs to distinguish between the general
phenomenon of transcendental illusion (that is, the appearance that D is true) and the
particular illusions to which it gives rise. Kants position appears to be that we are to think
of the particular illusions as resting on certain particular dialectical inferences, which
are their immediate grounds, but also that we are to think of these inferences themselves
as resting, in turn, on the general illusion that D is true.
465
I A N P R O O P S
dialectical inferences, which are set over against the Antinomies, on the
one hand, and the arguments of the Ideal of Pure Reason, on the other.
In the J asche Logic Kant divides formal fallacies into paralogisms
and sophisms as follows: a fallacy counts as a paralogism insofar as one
deceives oneself through it, but as a sophism insofar as one intentionally
seeks to deceive another (9:13435). This contrast, however, is unlikely
to be the one guiding Kants use of the term paralogism in the frst
Critique; for if it were, he would be implying that those dialectical infer-
ences of reason (A 309/B 366) that are discussed outside the Paralo-
gisms chapterthose found in the Antinomies and in the Ideal of Pure
Reason
29
are in fact merely sophisms. And yet it is scarcely credible that
Kant would have taken all of the unsound arguments criticized in these
other chapters to have been propounded deceitfully, even if he supposes
that some of them might have been.
30
Indeed, Kant seems to imply that
these arguments are often given in good faith. For example, speaking
quite generally of the dialectical witnesses that transcendent reason
brings forth on behalf of its pretensionsand so, presumably, of the
arguments of the Antinomies and Ideal among others, he says: we knew
beforehand that all their allegations, while perhaps honestly meant, had to
be absolutely null and void. (A 703/B 731, emphasis added). For these
reasons, the paralogism-sophism contrast does not seem to be a good
guide to the thinking behind Kants decision in the frst Critique to style
a particular class of dialectical inferences paralogisms.
A better text for these purposes occurs at the beginning of the
Paralogisms chapter itself:
A logical paralogism consists in the falsity of a syllogism due to its form,
whatever its content may otherwise be. Atranscendental paralogism, how-
ever, has a transcendental ground for inferring falsely due to its form.
(A 341/B 399)
29. In the case of the Ideas of the world and God, Kant also takes transcendental illu-
sion to lead to certain dialectical inferences. He maintains that in the former case these
inferences take the form of hypothetical syllogisms, in the later, disjunctive syllogisms.
A 340/B 398; A 406/B 43233.
30. Kant does suggest that some of these arguments are deployed deceitfully. See,
for example, his discussion in the Hechsel Logic of the fallacy sophisma heterozeteseos (the
fallacy of searching for something else) as it occurs in the proof of the existence of God.
Kant suggests that in this case one pretends to give a proof of an interesting claim while
actually giving a proof of a claim that is much weaker and consequently less interesting.
Young 1992, 41011. These notes, which are based on lectures Kant gave in the early
1780s, were not discovered until the 1980s and so are not included in Kant 1902. For a
discussion of their dating, see Pinder 1987.
466
Kants First Paralogism
The part of this passage that deals with logical paralogisms is echoed both
by the Blomberg Logic, where Kant says that Every inference false in form
is a paralogismus (24:287), and by a marginal note to the frst edition of
the frst Critique, where Kant says that A paralogism is a syllogism that is
false in forma (Erdmann 1881, 154; 23:38, quoted in Guyer and Wood
1998, 411). Clearly, then, a paralogism is a formally invalid syllogism. But
there is more to it than that, as a refection from the Duisburg Nachla
makes clear:
A paralogism is a syllogistic inference that is false as far as its form is con-
cerned, although as far as its matter (the antecedent propositions) are
concerned [bei Vorders atzen], it is correct. (R 5552; 18:218, translation my
own
31
)
So a paralogism is false in form,
32
but correct in matter.
33
This remark
suggests that what Kant means by a paralogism in the frst Critique is
31. In rendering the dative plural phrase bei Vorders atzen as the antecedent propo-
sitions, I follow the practice of J. Michael Young. See his translation of the J asche Logic,
especially sec. 59 (9:121) in Young 1992. It should be noted that the translators of the
Cambridge edition of Kants refections are misleading here. They render this plural
noun as the major premise (Guyer 2005, 236), thereby giving the false impression that
Kant once supposed that the major premise alone constituted the matter of a syllogistic
inference. Kant, however, always supposed that both premises constituted the matter of
the argument.
32. In the Hechsel Logic Kant glosses an arguments being false as to form as its being
a fallacy (16:110); so there is reason to think that, in characterizing a paralogism as
false due to its form in the frst Critique (A 441/B 399), Kant is simply characterizing it
as formally fallacious. This apparently obvious point is worth emphasizing because Kant
elsewhere equates an arguments form with its conclusion (for example, J asche Logic, sec.
59). However, if Kant really thought we knew a paralogisms conclusion to be false, he
would be dogmatically committed to its negation. His claim, for example, that the frst
paralogism is false due to its form would be tantamount to the claim that the self is
not a substance
1
. To avoid saddling Kant with such an implausibly dogmatic stance, we
need to allow ourselves to be guided on this point by Hechsels notes in preference to
J asches and to understand a paralogisms form as what we would today call its logical
form rather than its conclusion.
33. A historical precedent for treating the premises of an argument as its matter
and its logical form as its form occurs in the writings of Christian Wolff: with infer-
ences one is to look in part to the correctness of the propositions, in part to their mode
of connection with one another, that is, as one usually says for short, in part to the mat-
ter, in part to the form. Wolff 1747, sec. 390, translation my own. It is also present in
St. Thomas Aquinass commentary on Aristotles Physics: Each argument of both Melis-
sus and Parmenides is sophistic, for they err in matter, whence [Aristotle] says that they
have accepted what is false, that is, they assume false propositions, and they err in form,
whence he says they are not syllogizing. Aquinas 1999, 10. This kind of application of the
matter/form distinction to arguments is absent from the passages of Aristotle on which
St. Thomas is commentating here.
467
I A N P R O O P S
an inference that is invalid, even though its premises are true. (We shall
soon see that there is confrming evidence for this interpretation in the
Paralogisms chapter itself; but it will be convenient to defer its discussion
until the next section.) If this idea is correct, it would serve to distinguish
the Paralogisms from the arguments of the Antinomies and the Ideal
since, among these three groups of dialectical inferences, only the Paral-
ogisms are such that every member of the group is taken to be an invalid
argument with true premises.
It is worth noting, further, that each paralogism is supposed to
involve a fallacy of equivocation. That much is clear from the continua-
tion of the refection from the Duisburg Nachla:
[A paralogism] arises when the middle concept is taken in different
senses in the two premiseswhen, namely, the logical relation (in think-
ing) in one of the premises is taken as a real one (of the objects of intu-
ition) in the other. (R 5552; 18:218)
What this refection proceeds to discuss are not fallacious inferences
about the self but rather the body of fallacious inferences fromthe mere
actions of refection (A 278/B 334) that Kant terms the amphiboly of
the concepts of refection. And although his deletions indicate that he
later revised his view that the concepts of refection lead to paralogisms
(R 5552; 18:218), the fact that he once considered applying the concept
of a paralogism to subject-matter unconnected with the self suggests that
paralogisms are arguments of the kind just described, whether or not
they concern the self.
As we might have expected given that Kants focus in the Paralo-
gisms chapter lies specifcally on transcendental paralogisms, he is at pains
to emphasize the naturalness and inevitability of our tendency to want to
engage in paralogistic reasoning. Speaking of dialectical inferences in
general (and so of the transcendental paralogisms among other things),
he says:
They are sophistries not of human beings but of pure reason itself, and
even the wisest of all human beings cannot get free of them; perhaps
after much effort he may guard himself from error, but he can never
be wholly rid of the illusion, which ceaselessly teases and mocks him.
(A 339/B 397)
The paralogisms lead us to believe that we can have some substantive
(that is, synthetic) knowledge of the self as it is in itself. In truth, however:
468
Kants First Paralogism
The subject of inherence [that in which thoughts inhere] is desig-
nated only transcendentally through the I that is attached [angeh angte]
34
to thoughts, without [our] noting the least property of it, or being
acquainted with it or knowing it [through propositions] at all. (A 355,
translation my own
35
)
Kant is saying both that we cannot intuit the self as it is in itself and that
we cannot attain to propositional knowledge of itand presumably he
means propositional knowledge of its inner character or nature; for he had
earlier claimed that the self is cognized [erkannt] through the thoughts
that are its predicates (A 346/B 404, translation my own). Thus, for
example, while I can know that I am thinking that its raining, I cannot
know whether or not the subject of this thought is a substance
1
.
4. The Paralogistic Fallacy in the B-edition
Kant gives different accounts of the dialectical inference constituting the
frst paralogism in the two editions of the frst Critique. I will focus on
the second editions version because it belongs to a chapter that Kant
rewrote and which he expressly recommends as an improvement on the
text it replaces (see B xxxviii; compare Prolegomena, 4:381). I will, how-
ever, briefy discuss the A-edition version in the next section.
Precisely how is the formal fallacy of the B-edition paralogism to
be explained? Kants interpreters are far from agreement on this ques-
tion, and, as it seems to me, none has yet arrived at a wholly satisfactory
34. Here I followKemp Smith (1933) in favoring the looser attached to the more lit-
eral appended since the I think is typically prefxed to thoughts rather than appended
to them.
35. My rendering of this passage diverges from several established translations in
its treatment of the troublesome phrase ohne . . . uberhaupt etwas von ihm zu kennen,
oder zu wissen. Guyer and Wood offer without . . . cognizing or knowing anything at
all about it. But according to their own principles of uniform translation (Guyer and
Wood 1998, 73) cognizing would be appropriate only if the corresponding verb was
erkennen. Kemp Smith (1933) offers without knowing anything of it either by direct
acquaintance or otherwise. However, as Kitcher notes, this both misleadingly deploys
the Russellian term of art, direct acquaintance and fails to bring out the nature of
whatever notion it is with which kennen is being contrasted. Kitcher 1982, 528. Kitcher
herself renders the sentence in which the phrase occurs [we have no knowledge of the
thing that thinks] through awareness or through reasoning. Ibid. However wissen car-
ries no implication of reasoning but rather, through the contrast with kennen, one of
having propositional knowledge. My own translation resorts to an interpolation to bring
out the (familiar enough) contrast between kennen and wissen as relating to knowl-
edge of/acquaintance with, on the one hand, and knowledge about, on the other.
469
I A N P R O O P S
answer. Before offering my own account, it will be useful to state some
constraints on any adequate interpretation.
First, as our consideration of Kants use of the word paralogism
has suggested, any textually faithful reading will need to treat the frst
paralogismas a fallacious categorical syllogismwithanambiguous middle
term. That this constraint is operative in the frst paralogismis confrmed
by Kants remark that the conclusion of the frst paralogism is drawn per
sophisma fgurae dictionis (A 402 and B 411), which, as he explains in his
logic lectures, means that the paralogism has an ambiguous middle term
( J asche Logic, sec. 90, note; 9:135).
Second, as our refections in the preceding section also suggested,
any adequate reading must portray the paralogisms premises as true.
This conclusion is confrmed by Kants remark in the A-edition: If one
wants to give a logical title to the paralogism in the dialectical syllogisms
of the rational doctrine of the soul, insofar as they have correct premises, then
it can count as sophisma fgurae dictionis (A 402, frst emphasis added).
Since in the B-edition Kant again says that the frst paralogisms conclu-
sion is drawn per Sophisma fgurae dictionis (B 411), it seems likely that by
his lights the B-edition version of the argument also contains correct
(that is, true) premises.
A third constraint on any adequate reading is that it should por-
tray the fallacious inference as highly seductive. What transcendental
illusion produces in us, after all, is not a compulsion to err, but merely
a temptation. And this temptation cannot be adjudged so forceful that
it could induce someone to accept a patent non sequitur as a valid
argument.
With these constraints in hand, we are fnally in a position to ana-
lyze the frst paralogism. In the B-edition it is presented as consisting in
the following categorical syllogism:
What cannot be thought otherwise than as subject does not exist other-
wise than as subject, and therefore is substance.
Now a thinking being, considered merely as such, cannot be thought oth-
erwise than as subject.
Therefore it also exists only as such a thing, that is, as substance. (B 410
11)
This argument gives the misleading impression of being an instance of
a valid argument-schema, namely, the frst-fgure categorical syllogism in
Barbara:
470
Kants First Paralogism
All M are Pmajor premise
All S are Mminor premise
Therefore: All S are Pconclusion
The B-paralogismor, rather, a slightly more regimented (and clarifed)
version of itis obtained by replacing the schematic letters as follows:
M: entities that cannot be thought otherwise than as subjects
P : entities that cannot exist otherwise than as subjects, and therefore (by
defnition) are substances
S: entities that are thinking beings (considered merely as such)
Two remarks about this regimented version of the paralogism are in
order. First, I have supplied the term entities, which is intended to
be categorially neutral, merely in order to massage the argument Kant
actually presents into the (apparent) form of a frst-fgure categorical syl-
logism in Barbara. This requires construing each of the predicates as a
plural noun-phrase. Second, in order to be consistent with Kants def-
nitions of substance at B 149, B 288, and B 289, the predicate-term, P,
needs to read entities that cannot exist otherwise than as subjects . . .
rather than entities that do not exist otherwise than as subjects . . . Thus
our formulation of P involves a modal strengthening of what Kant actu-
ally writes. The need for this stronger formulation is further indicated by
the footnote of B 41112, in which Kant presents a consequence of the
arguments intended (but not attained) conclusionas being that I cannot
exist otherwise than as a subject (emphasis added). In what follows I will
treat Kants omission of the needed modal force in his main statement
of the argument as merely a slip.
As we have noted, the argument does not actually have the formof
a frst-fgure categorical syllogismin Barbara on account of the ambiguity
of its middle term (A 402). Kant locates the source of this ambiguity in
the word thought:
Thought [Das Denken] is taken in an entirely different meaning [Bedeu-
tung] in the two premises: in the major premise, as it applies to an object
in general (hence as it may be given in intuition); but in the minor
premise only as it subsists in relation to self-consciousness, where, there-
fore, no object is thought, but only the relation to oneself as subject (as
the form of thinking) is represented. (footnote, B 41112, translation my
own)
The interpreters task is to identify these two senses of thought. We will
aim to do so by considering the intent of each of the premises.
471
I A N P R O O P S
The major premise says, in effect, that if it is impossible to conceive
of some entity as a property or mode, then that entity cannot exist as a
property or mode. Since the premises of the paralogism are supposed to
be true, Kant must agree with this generalized conditional. He nonethe-
less avoids dogmatism because he also insists that we can never know any
instance of the antecedent to be true.
36
As he sees it, the conclusion that
something cannot be conceived of as merely a property or mode of some-
thing else must always be premature since it amounts to the dogmatic
assertion that that thing is an ultimate subject.
37
Kant, then, has a general reason for denying that the predicate
fguring in the antecedent of the major premisecannot be thought
otherwise than as subjectcan be known to apply to anything. But, in
addition, he has a specifc reason for denying that this predicate can be
known to hold of a thinking being considered merely as such. For it is at
least arguable that the selfto take one such beingcan be conceived
otherwise than as a subject (that is, as a property or mode of some-
thing else). Indeed, one version of such a view is standardly attributed to
Spinoza, who holds the human mind to be a mode inhering in the one
true substance: God.
38
Spinoza may or may not be right about this, but
his position is at the very least not known to be unintelligibleand the
same can be said for the alternative ideas that the self is merely a power of
a thing,
39
and that it is a confguration of other things. But such facts are
36. One might reasonably doubt whether by the lights of his own epistemology Kant
could really be entitled to grant the truth of the major premise. For it is not clear upon
what grounds he could take himself to know that such a claim is true. Since the issue is
rather involved, I reserve its discussion for section 6 below, Objections and Replies.
37. Compare a refection conjecturally dated to 177375: Our reason is not a faculty
the progress of whose use we can arbitrarily cut off. . . . It is contrary to reason to cut off
further research and wantonly suspend all further effort. R 4741; 17:694.
38. In the Ethics Spinoza says that the human mind is part of the infnite intellect
of God. Ethics 2, P 11, Corollary; Spinoza 1992, 70. Everything for Spinoza is a mode or
a substance. Ethics 1, P 15, Proof; Spinoza 1992, 40. Since God is the sole substance, the
self is not a substance and so must be a mode.
39. Kant was unquestionably aware of one view according to which the soul is not a
substance
1
. This is Baumgartens viewaccording to whichthe soul is to be identifed witha
representational power, namely, the power of representing the universe froma particular
point of view. See Baumgarten 1902, sec. 741; and see 29:906 for Kants awareness that
Baumgarten held such a view. As Karl Ameriks notes (2000, 82, n. 94), Kant attributes
a similar view to Wolff (28:261), and he takes it to imply that the soul is an accident.
28:261. However, Kant gives no citation when he makes this attribution, and one has to
wonder whether Wolff ever held such a view. Certainly, if he did, he was not consistent in
doing so; for in Wolff 1747 he clearly maintains that the soul is a self-subsisting thing that
has powers (see secs. 74344). What matters for our purposes, however, is that although
472
Kants First Paralogism
all Kant requires in order to argue that the minor premise, interpreted
as it would need to be in order to mesh with the major, is not known
to be true. So, even if the argument were interpreted so as to be valid,
it would still not be known to be sound, and so would not constitute a
demonstration.
The major premise, I contend, involves treating thought as
meaning conception. I take Kants position to be that in the minor
premise thought does not have this sense, even though, when in the
grip of transcendental illusion, we are tempted to think that it does. But
then what sense does it have?
The best place to begin the examination of this question is with
the continuation of the footnote from B 41112. Kant says:
In the frst premise, things are talked about that cannot be thought of
other than as subjects; the second premise, however, talks not about
things but about thought (in that one abstracts from every object), in
which the I always serves as subject of consciousness; hence in the con-
clusion it cannot follow that I cannot exist otherwise than as subject, but
rather only that, in thinking my existence, I can use myself only as the
subject of the judgment, which is an identical proposition that discloses
absolutely nothing about the manner of my existence. (footnote B 411
B 412, translation my own)
Let us focus on the conclusion about the self that Kant supposes we are
entitled to draw from the premises of the frst paralogism, namely:
[1] In thinking my existence I can use myself only as the subject of
the judgment.
Because Kant supposes we are entitled to infer [1] from the premises of
the argument, and because he takes those premises to be true, he must
also take [1] to be true. So a charitable interpretation of [1] is in order.
But any charitable reading of this claim must take Kant to be committing
an error that is the analogue at the level of judgment of a confusion of
use and mention. What I use as the subject of a judgment expressing
my existence, after all, is not my self but only a judgmental component
representing my self.
40
So Kant must mean:
Kant takes issue with the view that the soul is a power, he does not seem to regard it as
unintelligible. See 28:26162 and 29:906.
40. Kants talk of using the self or the I can be clarifed by comparing it to his
discussion of the use of more ordinary concepts in judgments. In the Metaphysical Foun-
dations of Natural Science, inthe course of explaining the notions of substance and accident
(understood as pure concepts of the understanding), he says:
473
I A N P R O O P S
[2] In thinking my existence I can use the representation of my self
only as the subject of the judgment.
This is better, but we are not yet fnished with our charitable reformula-
tions; for [2], as it stands, is, strictly speaking, false. To see why consider
the judgment: Something instantiates the property of being identical
with me. In this judgment I think my existence without using the rep-
resentation of my self as a subject, for it is used only as a component of
the predicate. Nor do I use the representation of my self as a subject in
the judgment: The property of being me is exemplifed, for here my
representation, me, is used only as a component of the subject con-
cept. To do justice to Kants thought, therefore, we will have to suppose
that when he says such things as [1] he does so against the background
of the assumption that any subsentential component of a categorical
judgmental occurs either as a subject or as a predicate.
41
Although this
assumption is, strictly speaking, false, it does little harm to Kants rea-
soning. To correct for it, we need only suppose that, when he says that a
In the categorical judgment The stone is hard, the stone is used as subject, and hard
as predicate, in such a way that the understanding is still free to exchange the log-
ical function of these concepts, and to say that something hard is a stone. (4:475,
continuation of note

)
Here when Kant says the stone is used as subject it is clear that he means that the
concept stone is used as subject. The emphasis on the word stone thus seems intended
to indicate that in this context this word does not refer to an object (a stone) or even to
stones in general (the plurality of stones) but only to a concept. The same typographical
convention seems to be in play in Kants use of the phrase the I at A 349 when he says:
in all our thinking the I is the subject, in which thoughts inhere only as determinations,
and this I cannot be used as the determination of another thing. The parallel with the
phrase the stone suggests that the emphasis on the word I in the phrases the I
and this Iindicated in the Akademie edition by spaced typesettingis intended to
indicate that Kant means to be referring to the judgmental component expressed by the
word Iwhich component is presumably the concept of the soul spoken of at A 351.
His point, therefore, is that the concept expressed by the word I cannot be employed in
a judgment in a predicative role. Note: some editions of the frst Critique, including the
Guyer-Wood translation, depart from the Akademie edition by omitting the emphasis on
I in the phrase this I while including the emphasis in the phrase the I. For all I know,
this may be because they are relying on a text that is (at this point) more authoritative
than the Akademie edition. But even if that should be so, the fact that the phrase this I
is anaphoric on the phrase the I strongly suggests that Kant would have intended the
emphasis to be present in both phrases.
41. Kant does not treat the copula as expressing a judgmental component. Instead,
he treats it as expressing the judgments form. See, for example, J asche Logic, sec. 24,
(9:105).
474
Kants First Paralogism
judgmental component can be used only as [the] subject of the judg-
ment, what he really means is that it cannot be used as [the] predicate
of the judgmentwhereby a predicate here he means a judgmental
component in predicative role, rather than a property or mode. At the
end of the day, then, what Kant must mean by [1] is just:
[3] In thinking my existence I cannot use the representation of my
self as [the] predicate of the judgment.
42
At this point it would be well to recall what Kant says about [3] in the
footnote to B 41112. He says that all that follows from the argument is
[3] in contradistinction to the more metaphysically loaded claim that I
cannot exist otherwise than as subject. The focus on existence in each
of these claims suggests that Kant means to be saying that:
[4] All that follows from the premises of the paralogism as far as my
existence is concerned is not that I cannot exist otherwise than as
subject but merely that [3].
It is easy to see why Kant should take [3] to follow from the premises of
the paralogism; for it follows from a combination of the minor premise,
construed as saying
[5] The representation of a thinking being, considered merely as
such, cannot be used as the predicate of a judgment,
with the independently plausible claim that
[6] The representation I (or me)or, strictly, the judgmental
component expressed by these wordsis the representation of
a thinking being considered merely as such.
There is, then, good reasonprovided by footnote B 41112to
take the minor premise as meaning [5]. And, as we required, the minor
premise, so construed, seems to be true. For the representation of any
thinking being considered merely as such will be a singular represen-
42. Kant describes [1] as an identical proposition, which means that it is an analytic
judgment. A7/B10; J asche Logic, 9:111. One supposes that he must viewit this way because
it (a) contains necessity but (b) is not plausibly construed as a claim whose truth is a
condition of the possibility of experience. One supposes that Kant would have regarded
[1] as analytic because it amounts to [3] and because it is analytic of our concept of
the word Ich that it has the grammatical distribution it does. Presumably, it would be
thought of as That German word whose phonetic signal (in such-and-such a dialect)
is so-and-so and whose meaning is such-and-such and whose grammatical distribution is
so-and so, where this last so-and-so would contain the component does not occur as a
predicate. Such an account might be satisfactory for the word Ich, though one has to
worry whether it would really carry over to the level of judgmental components.
475
I A N P R O O P S
tation (whether it be a proper name, an indexical, a demonstrative, or
a defnite description). And representations in this category cannot play
the grammatical role of predicates
43
though, of course, they may fgure
as parts of predicates. Inthe case of a termfor a universal, by contrast, that
term can occupy either subject or predicate position in a judgment. For
example, both Blue is a color and American mailboxes are blue are
grammatical. (Of course, the same goes mutatis mutandis for what really
concerns us: the judgmental components these words express. Hereafter,
this point will be left unstated.)
Kants key observation about the B-paralogismthe point on
which his entire critical analysis of the argument turnsis that this fact
about the grammatical distribution of the frst-person singular personal
pronoun entails nothing about how the self can or cannot be conceived.
A fortiori, it entails nothing about whether or not the self is a substance
1
.
Thus, although the representation of the self cannot be used as the pred-
icate of a judgment, this fact discloses absolutely nothing about the man-
ner of my existence (B 412). This central point can be underscored by a
comparison with the case of an ordinary proper name. Suppose I were to
name my right fst Fred. Kant would accept that I cannot put the word
Fred in predicate position; but this emphatically does not show that
Fred is a substance
1
. Indeed, since Fred is something whose existence
depends on the confguration of other things, it must be deemed a mode;
hence, a determination of other things.
On this reading, the ambiguity on which the paralogism trades
can, as Kant suggests, be located in the word thought. In the major
premise to think of something is to conceive of it. But in the minor
premise to think of something is to assign its representation a certain
role in a judgment. Thus, in the sense appropriate to the major premise,
an entity that cannot be thought as a predicate is something that cannot
be conceived as a property or mode of another thing; while in the sense
appropriate to the minor premise an entity that cannot be thought as a
43. One might be inclined to doubt this on the grounds that in the judgment, for
example, that Allah is God, it looks as though God occupies predicate position. How-
ever, inspite of appearances to the contrary, this judgment does not in fact containa pred-
icate position since it is not of subject-predicate form. Rather, it is an identity judgment,
and so of relational form. In a Kantian setting, of course, this point may be less obvious
than it is to us today since Kant failed offcially to recognize relational judgments as hav-
ing a distinctive logical form. Nonetheless, it remains the case that the fact that God is a
singular term disqualifes it from predicate-position, even if Kant himself lacked some of
the resources needed to defend this point against alleged counterexamples of this kind.
476
Kants First Paralogism
predicate is something whose representation cannot play the grammati-
cal role of a predicate. Exploiting the ambiguity of predicate, which in
Kants writings sometimes means a property (or mode) and sometimes
a judgmental component in predicative role, we might characterize the
ambiguity in question as that between the notion of something that can-
not be conceived as a predicate and the notion of something that cannot
be deployed as a predicate. Alternatively, sacrifcing elegance to precision,
we might say that the kernel of Kants critical insight amounts to the fol-
lowing observation: the fact that somethings representation cannot be
deployed as a predicate in judgments doesnt entail that that thing itself
cannot be conceived as a property or mode of another thing, and, a for-
tiori, it doesnt entail that it cannot exist inthis manner. That observation,
I take it, constitutes a philosophical insight of real depth.
44
44. Among the interpretations of the B-edition paralogismknown to me, the one that
comes closest to my own is that of Karl Ameriks (2000). According to Ameriks (2000, 71),
the reason there can be no sound argument of the kind the frst paralogism attempts to
be is that despite systematically misleading appearances, we dont have suffcient evi-
dence to make any subsumption under the frst premise. I agree with that assessment;
but my interpretation of the B-paralogism differs from Amerikss on one substantive
point and three points of emphasis. The substantive point is that the middle term (in
the formally presented argument) can (and, if we are to make sense of Kant, must) be
viewed as ambiguous in two places: in the word thought and in the word entities. The
frst point of emphasis is that the minor premise is not, in the frst instance, as Ameriks
would have it, a claim about how I must represent myself whenever I consider myself
merely as thinking (that is, as a subject). It is more specifcally a claim about where my
representationparadigmatically, the word Icanor cannot occur ina judgment. One
might, of course, offer this as a gloss on how I must represent myself whenever . . .;
so on this point, our readings are, I think, compatible. However, they differ insofar as
Ameriks does not in fact offer this gloss, and so leaves it somewhat unclear what it is to
represent the self as a subject. A second (very minor) difference of emphasis relates to
the idea that the focus of the minor premise lies not, as Ameriks would have it, on the
point that I must think of myself as subject, which, on my reading, would mean that the
representation I must occur in subject position; it lies rather on the point that the rep-
resentation I cannot go in predicate position. A third difference of emphasis relates
to my employment of the idea that, in relation to self-consciousness (that is, in the
epistemic state Im in at the terminus of Cartesian doubt), my resources for thinking of
myself are diminished in such a way that, under the infuence of transcendental illusion,
it will easily seem that I have nothing to appeal to but an inappropriate grammatical
test when considering whether or not I can conceive of myself as a determination. In
spite of these differences, I believe there is much to admire in Amerikss account, and, in
particular, I agree with many of his criticisms of other interpreters. See Ameriks 2000, 68,
7173.
477
I A N P R O O P S
5. The Paralogistic Fallacy in the A-edition
We are now equipped to attempt a reading of the A-edition version of the
frst paralogism.
45
I will argue that Kants criticism of this version of the
paralogism consists, at bottom, in the observation that it operates with
the wrong conception of substancewrong because it fails to corre-
spond to any notion the rational psychologist might have in mind when
claiming that the self is a substance. This conception turns out to be even
weaker than the notion of substance
1
and to make an essential appeal to
considerations having to do with where a judgmental component can
(and cannot) occur in a judgment.
Kant formulates the A-edition version of the argument as follows:
That the representationof whichis the absolute subject of our judgments,
and hence cannot be used as the determination of another thing, is sub-
stance.
I, as a thinking being, am the absolute subject of all my possible judg-
ments, and this representation of Myself cannot be used as the predicate
of any other thing.
Thus I, as thinking being (soul), am substance. (A 348)
This argument diverges even farther than did the B-edition version from
the strictly correct form of a categorical syllogism: the frst premise is a
defnition rather than a universal claim stating a suffcient condition for
being a substance; and the minor premise isnt general. Putting it into
the form of a syllogism with a singular minor premise and simplifying by
eliminating the redundant, defned predicate: is the absolute subject of
our judgments, we may reformulate the argument as follows:
Entities whose representation cannot be used as a determination of other
things are substances.
The self is an entity whose representation cannot be used as a predicate
of another thing.
Therefore the self is a substance.
On the assumption that determination in this context just means pred-
icate, the argument is valid. Moreover, because, as we have seen, the rep-
resentation of the self cannot be used as a predicate, the minor premise
is true. Since the major premise is just a consequence of a defnitional
truth, the argument is sound. The arguments only shortcoming is that
45. The ideas in this section have benefted greatly froma conversation with Kenneth
Winkler, who convinced me that the conclusion of the A-edition frst paralogism had to
be somehow weaker than the proposition that the soul is a substance
1
.
478
Kants First Paralogism
its conclusion is too weak to serve the rational psychologists purposes.
It establishes only that the self is a substance in the grammar-driven
sense of something whose representation cannot occur as a predicate in
a judgment. As Kant puts it, the argument establishes that the soul is a
substance only in the idea (A 351).
46
In order to emphasize that this
is the weakest of the conceptions of substance we have considered so
far, it will be convenient to label it substance
0
.
47
As Kant observes, the
argument doesnt entail anything metaphysically interestingleast of all
the souls natural immortality (A 35051; A 400); for, as weve seen from
our discussion of the case of Fred, some substances
0
can cease to exist
through the decomposition or rearrangement of their parts.
Because the A-edition frst paralogism employs a weak and meta-
physically neutral conception of substance, there is no need to posit a
fallacy of equivocation to diffuse its force. But, equally, lacking such an
equivocation, it fails to satisfy Kants offcial description of a paralogism.
48
46. More precisely, he says that the concept of the soul signifes a substance only in
the idea. This entails that the soul itself is a substance only in the ideaand that means
that the souls representationthe I of apperceptioncannot be used as a predicate
in judgments.
47. Kants notion of substance
0
seems not to have any close precedent in the tradi-
tion. It might be thought to correspond loosely to Aristotles notion of something that
is not said of a subject from the Categories (sec. 5) (Aristotle 1984); for its possible
to read Aristotle here as offering a characterization of substance that rests, partly, on
grammatical considerations. But that idea is only one element of Aristotles notion of
primary substance rather than the notion itself. It seems that Kant was led to the notion
of substance
0
by his attempt in the A-edition to analyze the paralogisms quite generally as
establishing nothing more than claims about the self in the idea or in concept. The
generality of this ambition is evident from his remark that mere apperception (I)
is substance in concept, simple in concept, etc., and thus all these psychological the-
orems are indisputably correct. Nevertheless, one by no means thereby cognizes any-
thing about the soul that one really wants to know, for all these predicates are not
valid of intuition at all, and therefore cannot have any consequences that could be
applied to objects of experience; hence they are completely empty. (A 400)
48. Karl Ameriks has suggestedquite plausiblythat Kants A-edition criticisms
of the frst paralogism are directed not at the argument Kant presents there as the
paralogistic syllogism per se but rather at an invalid extended argument for the perma-
nence of the soul. Ameriks 2000, 68. According to Ameriks, the rational psychologists
error consists in passing fromthe conclusion that I cannot be a determination of another
thing (that is, that I am a substance
1
) to the further conclusion that I am a substance
2
.
Such a step involves a version of the fallacy Jonathan Bennett calls infating the frst
paralogism. Bennett 1974, sec. 25. The B-edition improves on this diagnosis and criti-
cism by locating the fallacy squarely within the paralogistic syllogism itself. I agree with
the general style of this analysis, but I would claim that the conclusion that gets infated
479
I A N P R O O P S
Inorder to present the belief inthe substantiality of the soul as reached by
a genuine paralogism, what is needed is a better defnition of substance;
and that is precisely what the B-editions version of the paralogism works
withnamely, substance
1
.
As I see it, then, the main change between the A- and B-edition
versions of the frst paralogisma change that enables Kant to portray
the paralogism as invalidis the change in the defnition of substance
from that the representation of which cannot be employed as a determi-
nation of another thingsubstance
0
(compare A 348)to that which
cannot exist otherwise than as subject
49
substance
1
(B 149, B 288, B
289). This thought is confrmed by the absence of the latter notion from
the A-edition and by Kants omission of the former from the parts of the
B-edition that he rewrote. I would therefore claim that it is his reformu-
lation of the notion of substance that constitutes the main improvement
in his presentation of the frst paralogism.
6. Objections and Replies
The account I have given of the B-edition paralogism satisfes our main
desiderata: it portrays the paralogism as a fallacious syllogism contain-
ing an ambiguous middle term; the fallacy is suffciently subtle that it
might have been committed by thinkers of some sophistication; and the
equivocation is located precisely where Kant says it isnamely, in the
word thought. So the reading has its merits. But what are its points of
vulnerability?
One possible line of objection alleges that the interpretation
would commit Kant to predicting the occurrence of dogmatism where
it does not exist. If Kant is right that we are apt, when in the grip of
transcendental illusion, to confuse the idea that the selfs representation
cannot occur in predicate position in a judgment with the idea that the
self cannot be conceived as a determination (that is, as a property or
mode), why, one might want to know, arent we apt to make a similar
mistake in other cases? Why, for instance, shouldnt we also be inclined
to think of my right fst, whose representation, Fred, cannot occur in
predicate position, as a substance
1
?
(and which constitutes a lemma in the invalid, extended argument) is that the soul is a
substance
0
rather than, as Ameriks seems to suppose, a substance
1
.
49. Recall that I amtreating the omission of modal force in the predicate of the major
premise in the B-version as just a slip.
480
Kants First Paralogism
The reply, I take it, should run as follows: we are not tempted to
take Freds inability to occupy predicate position as a sign that what it
represents is incapable of being conceived as a determination of another
thing because what it represents is intuitable. In intuiting Fred, I can per-
ceive that it is complex and thereby come (perhaps tacitly) to understand
that it is something whose existence consists in the confguration of other
things. In other words, sensible intuition can assure me that Fred is a
mode. In the case of the self, by contrast, all I have to go on in consider-
ing whether or not it can be conceived of as a determination of another
thing are the features that could be known to attach to it at the terminus
of Cartesian doubt. But, crucially, among these features is nothing that
would indicate that the self is not a substance
1
. They include only the
general property of having thoughts, along with the particular proper-
ties of believing that p, doubting that q, willing that r , and so forth.
50
It is
because in considering the self we have nothing to go on beyond these
meager resources that in this case the fallacy is so easy to commit.
So much, then, for the frst line of possible criticism. A second
line challenges my attribution to Kant of the idea that the judgmental
component expressed by I cannot occur as a predicate. Some grounds
for this possible objection are mentioned by Patricia Kitcher (1982, 526).
Although she herself wants to attribute something like this claim to Kant
as part of her interpretation of the A-edition frst paralogism, she does
so only with reservations. Kant, she says, never says why the I cannot
occur as a predicate. Further he himself points out that any concept can
occur ineither the subject or the predicate positionina sentence (ibid.).
As support for the second of these claims, Kitcher cites the texts: A 349, B
12829, and A 24243/B 300301. Kitchers contention, if correct, would
pose a diffculty for my reading only if what she means by a concept is
a judgmental component in general. For the sake of argument, I shall
suppose that that is so.
The reply to this objection simply involves observing that the pas-
sages cited do not in fact support the conclusion Kitcher draws from
them. Consider A 349. Far from suggesting that any judgmental compo-
nent can play the role of a predicate, this passage actually suggests that
the judgmental component expressed by I cannot do so. It contains
50. As I noted earlier, it is clear that Kant takes us to know at least this much about the
self, for he says, Through this I, or He, or It (the thing), which thinks, nothing further
is represented than a transcendental subject of thoughts = x, which is cognized [erkannt]
only through the thoughts that are its predicates. A 346/B 404, emphasis added, translation
my own.
481
I A N P R O O P S
the claim that in all our thinking the I is the subject, in which thoughts
inhere only as determinations, and this I cannot be used as the determination
of another thing (A 349, emphasis added). If I am correct that when Kant
speaks of using the I he means employing the selfs representation in
a judgment, this would have to be taken to mean that the judgmental
component expressed by the word I cannot occur as a predicate in a
judgment. At any rate, there is no hint of a suggestion at A 349 that this
representation can function as a predicate. Nor, so far as I can see, is
there any hint of this idea at A 24243/B 300301. The only text cited
by Kitcher that might seem to support the attribution of such a view is
the passage from B 12829. Here Kant offers a characterization of the
categories:
[The categories] are concepts of an object in general, by means of which
its intuition is regarded as determined with regard to one of the logical
functions for judgments. Thus, the function of the categorical judgment
was that of the relationship of the subject to the predicate, e.g., All bodies
are divisible. Yet in regard to the merely logical use of the understanding
it would remain undetermined which of these two concepts will be given
the function of the subject and which will be given that of the predicate.
For one can also say: Something divisible is a body. Through the cate-
gory of substance, however, if I bring the concept of a body under it, it
is determined that its empirical intuition in experience must always be
considered as subject, never as mere predicate; and likewise with all the
other categories. (A 94/B 12829)
As I read this passage, Kant is not claiming that any representation what-
soever can occur as a predicate in a judgment. Rather, he is claiming that
prior to the application of the category of substance, any general term
might play either subject- or predicate-role in a subject-predicate judg-
ment. As we have already noted (with our example of blue), this fexi-
bility is typical of general terms. That idea, however, is far from entailing
that I can occur in predicate position, for I is a representation that
purports to stand for a particular .
A third possible objection would charge that my interpretation
is fawed because it is incomplete. The problem is that immediately after
presenting the frst paralogism, Kant offers an explanation of the infer-
ences deceptiveness that appears to differ from the one in the footnote
on which we have been focusing so far:
The major premise talks about a being that can be thought of in every
respect, and consequently even as it might be given in intuition. But the
482
Kants First Paralogism
minor premise talks about this being only insofar as it is considered as
subject, relative only to thinking and the unity of consciousness, but not
at the same time in relation to the intuition through which it is given as
an object for thinking. (B 411)
The present objection simply charges that we havent yet done justice
to this apparently rather different explanation of the equivocation. The
reply to this objectionsuch as it isis just to concede its correctness
and to show that the needed further explanation is compatible with the
one given so far.
Lets begin by considering the contrast in the second sentence.
Kant says that the minor premise talks about a beingnamely, the self
in a way that contrasts with thinking of it in relation to the intuition
through which it is given as an object for thinking. Kant repeatedly
insists that the self as it is in itself is not an object of intuition; so he
cannot mean that the minor premise neglects to consider the self rel-
ative to the intuition through which it is given as an object for think-
ing. What he must mean, rather, is just that if, contrary to fact, the
minor and major premises were to mesh into a valid argument, the self
would have to be the kind of thing that could be given in intuition. He
might have said that the minor premise speaks only of entities that are
related to the unity of consciousness, not of entities that might be given as
objects.
Another reason, then, why the argument must be recognized as
invalid is that the major premise speaks of things that have the right onto-
logical status to be intuitedthat is, things that can be given as objects
while the minor premise speaks of nothing of this kind. This does not
bring out an ambiguity in the actual wording of Kants formulation of
the paralogism, but, applying a little charity, we might see the envisaged
ambiguity as being located in the word entities as it fgures in the more
formally correct formulation of the argument given in section 4 above.
Recall that this argument runs:
All entities that cannot be thought otherwise than as subjects are entities
that cannot exist otherwise than as subjects, and therefore (by defnition)
are substances.
All entities that are thinking beings (considered merely as such) are enti-
ties that cannot be thought otherwise than as subjects.
Therefore,
483
I A N P R O O P S
All entities that are thinking beings (considered merely as such) are enti-
ties that cannot exist otherwise than as subjects, and therefore (by def-
nition) are substances.
According to the line I am suggesting, we should say that in the
major premise entities is understood to mean objects, while in the
minor it is understood to mean things of another kindsubjective uni-
ties, perhapsunderstood as the pseudo-objects imagined to account
for the unity attaching to each persons thoughts in virtue of those
thoughts belonging to one consciousness.
There is no avoiding the conclusionthat Kant effectively offers two
complementary accounts of where the middle termthe whole phrase:
entities that cannot be thought otherwise than as subjectsis infected
with ambiguity, namely, in the word thought and in the word enti-
ties. But because he does not himself formulate the argument in its
strictly correct form, the term entities does not occur in the formu-
lation he offers. That being so, the account of the footnote must be
adjudged the more satisfactory of the two, and one can suppose that
Kant added the footnote in order to remedy this defciency in the main
text.
51
The fourth objection raises a problemfor our claimthat the major
premise of the B-edition paralogism would be true by Kants lights.
52
I
have argued that Kant understood the term paralogism in such a way
that the idea that the premises should be true follows trivially from the
very notion of a paralogism. However, one might reasonably wonder how
51. This account of the formal fallacy in the B-edition paralogism differs markedly
from Griers, which I fnd to be less successful than her account of transcendental illu-
sion. Grier suggests that the argument should be reconstructed as the invalid inference:
x (Ox (Sx Ex)), Sa; so Ea, where Ox is x is an object of possible experience;
Sx is x cannot be thought otherwise than as subject, and Ex is x does not exist oth-
erwise than as subject. Grier 2001, 163. Grier doesnt indicate what she takes a to be,
but one supposes it must be the singular termthe self, or, perhaps, the thinking subject
(considered merely as such). This reconstruction has several drawbacks. First, and most
obviously, it fails to portray the paralogism as a syllogism. Second, since it portrays the
minor premise as a singular judgment, it would seem to ft the frst-edition version better
than the second. Third, the very suggestion that the frst premise incorporates a restric-
tion that is absent from the second seems to run counter to Kants words. He describes
the frst premise as talking of a being that can be thought of in every respect and the
second as speaking of this being only relative to thinking and the unity of consciousness.
This suggests that the second premise is in fact more restricted than the frst.
52. I am grateful to an anonymous referee for pressing this question and to Kenneth
Winkler for discussion of some possible responses. The response I fnally settled upon
was, in its essentials, suggested to me by Ivan Heyman.
484
Kants First Paralogism
Kant could have taken himself to know the major premise to be true; for
while it seems to be intended to be read as a necessary claim, it is, at frst
sight, neither analytic nor synthetic a priori. I believe this objection can
be answered by arguing that Kant would have viewed the major premise
as analytic. Since this is far from obvious, however, it will be worth going
into this matter in some detail.
Consider the core of the major premise, namely: What(ever)
cannot be thought otherwise than as a subject does not exist otherwise
than as a subject. I argued earlier that in order to be consistent, Kant
ought to replace does not with cannot. Applying this emendation
and replacing the phrase otherwise than as a subject with the (in this
context) equivalent phrase as a determination, we obtain: Whatever
cannot be thought as a determination cannot exist as a determination.
To appreciate why Kant should have regarded such a claim as analytic it
is necessary to attend to the modalities involved. For each occurrence of
cannot, we must ask whether the modality in question is real or logical
impossibility. The answer, I believe, is that the frst impossibility is logical,
the second real. Thus the claim should be read: Whatever is such that
there is always a contradiction in representing it as a determination
of another thing cannot, as a matter of real possibility, exist as a
determination of another thing. Such a claim is analytic because it is
analytic of the notion of real possibility that whatever is really possible
is logically possible, so that whatever is logically impossible is really
impossible (compare 29:811). (We might say that x is really possible just
in case x is logically possible and x involves no real repugnancy.)
In support of such a reading of the relevant modalities, we may
observe that sometimes when Kant speaks of what can be thought he
clearly has inmind what canbe represented as logically possible. So much
is clear from, for example, Kants remark in the B-edition preface to
the frst Critique that [Although I cannot cognize freedom as a prop-
erty of my soul], nevertheless, I can think freedom to myself, that is, the
representation of it at least contains no contradiction in itself (B xxviii).
This provides some support for the idea that the frst impossibility in
the major premise is logical impossibility. (Or, more carefully, it suggests
that the whole phrase whatever cannot be thought otherwise than as
subject means whatever is such that there is always a contradiction in
representing it as a determination of another thing.) That the second
impossibility is real is only to be expected since the notion of impossi-
bility that features in the defnition of substance (as that which cannot
exist as a determination of another thing) is plausibly real impossibility.
485
I A N P R O O P S
Something, after all, would seemto still count as a substance
1
if the obsta-
cle to its existing as a determination of another thing were merely real
repugnancy rather than logical inconsistency.
Since the major premise is plausibly interpreted as analytic, we can
suppose that Kant would have taken it to be true. We can also suppose
that he would have taken the rational psychologist to have regarded it
as true.
7. A Closer Look at Transcendental Illusion
Before closing it will be useful to address some apparent problems for
applying to the Paralogisms chapter the account of transcendental illu-
sionoutlined inthe opening sections of this essay. That account, I believe,
is broadly correct; but it requires some further elaboration and refne-
ment. Two problems need to be addressed: First, although there is much
in the frst Critique to support the idea that the illusion that tempts us
to commit the fallacy of the frst paralogism is the illusion that Dor
rather an instance of it
53
is true, there are also numerous passages that,
on the face of it, suggest a rather different story. Second, the picture we
found helpful from the Prolegomenanamely, that of a regressive series
of conditions whose fnal member appears to be an ultimate subjectis
expressly repudiated at one point in the frst Critique. The present section
is devoted to showing that, in spite of these problems, the essential ele-
ments of the account of transcendental illusion given so far do nonethe-
less apply to the frst paralogism.
In the course of defending the completeness of his table of cos-
mological ideas (stated at B 443) Kant claims:
The category of substance and its accidents is not suited to a transcen-
dental idea, that is, in regard to this category reason has no ground to
proceed regressively toward conditions. (A 414/B 441)
54
53. The instance is: If a predicate is given, the whole series of subjects subordinated to
one another, a series which is therefore itself unconditioned, is likewise given.
54. Kant goes onto concede that we do have one concept of substance that might still
seem to be an idea of transcendental reason, namely, the concept of the substantial.
In this context, he understands this to be the concept of a subsisting object that lacks any
predicates. A 414/B 441. But since this would not, if it did exist, be an appearance (that
is, an object of possible experience), Kant thinks it plain that it cannot be thought of as a
fnal member in a regressive series of appearances, which is what would be required in the
present context. Note, incidentally, that we may think of the substantial as yet a fourth
conception of substance recognized by Kant.
486
Kants First Paralogism
Does this meanthat the Prolegomenas story is only a temporary aberration,
a view held feetingly between the two editions of the frst Critique? This
seems doubtful since the B-edition retains a number of remarks from
the frst edition in which the Prolegomenas account is plausibly present.
These remarks all occur in the chapter of the frst Critique dealing with
the general notion of a transcendental idea.
First, Kant describes the totality of conditions and the uncondi-
tioned as the common title of all concepts of reason (A 324/B 380).
Since one of these concepts of reason is the transcendental idea of the
self, this idea would seem to have as part of its content the notion of an
unconditioned totality of conditions. Second, in the course of arguing
that there are just as many species of deceptive syllogistic inference as
there are ideas, Kant says that in each species of [dialectical] syllogism
prosyllogisms proceed to the unconditioned: one, to a subject that is no
longer a predicate, another to a presupposition that presupposes nothing
further (A 323/B 37980, emphasis added). Here an absolute subject is
cited as an unconditioned, and the idea of a series of conditions is sug-
gested by the idea of prosyllogisms proceeding to the unconditioned.
Third, in the course of introducing the system of the transcendental
ideas, Kant explains that all transcendental ideas will be brought under
three classes, of which the frst contains the absolute (unconditioned)
unity of the thinking subject (A 334/B 391). If the thinking subjects
unity is unconditioned, it sounds as though its unity might be the stopping
point of a regress of conditions.
Together these remarks create the impression that in both edi-
tions of the frst Critique Kant takes seriously at least some parts of the Pro-
legomenas picture of the self as the unconditioned ultimate condition of a
regressive series. The passage from the Antinomies chapter also suggests
that he takes some of this picture back. So which parts of the Prolegomenas
story are supposed to go, and which are supposed to stay? It is not easy
to say, but a clue is provided by the prominent position accorded by both
editions to the following passage:
There will be as many concepts of reason as there are species of relation
represented by the understanding by means of the categories; and so we
must seek an unconditioned, frst, for the categorical synthesis in a sub-
ject, second for the hypothetical synthesis of the members of a series, and
third for the disjunctive synthesis of the parts in a system. (A323/B380)
55
55. Compare a partial quotation from refection 5553:
487
I A N P R O O P S
This remark constitutes the fourth paragraph of the section dealing
with the transcendental ideas. It thus occurs in a part of the book in
which Kant is laying out the grosser features of the Dialectics architec-
tonic. Here Kant is trying to forge a connection between, on the one
hand, the various kinds of synthesis represented by the ideas and, on the
other, the three species of dialectical syllogismcategorical, hypotheti-
cal, disjunctivehe takes to be associated with them. The three kinds of
unconditioned under discussion here are intended to match up with
the transcendental ideas of the self, the world, and God.
Kants words at A 323/B 380 suggest that he conceives of the tran-
scendental idea of the self as the idea of an unconditioned associated
specifcally with categorical synthesis. But now he implies that this uncon-
ditioned is not to be thought of as the frst member of a series. Instead,
that notion is reserved for the unconditioned as it relates to hypothetical
synthesis, a form of synthesis that Kant elsewhere implies is associated
with the Antinomies (A 406/B 43233). Given the prominence of this
remark, it seems wisest to assume that Kants more considered strand of
thought involves the rejection of the Prolegomenas picture of the self as
the fnal member of a regressive series. In the Antinomies Kant explains
the reason for this rejection as follows: Accidents (insofar as they inhere
in a single substance) are coordinated with one another, and do not con-
stitute a series (A 414/B 441).
So, to answer our question about the application of Kants model
of transcendental illusion to the frst paralogism: what goes (or, at any
rate, should go) is the idea of a series of conditions, while what stays (or
should stay) is the idea of a subject that is not (and cannot be) a predicate
of something else, and is, in that sense, unconditioned. Admittedly, Kant
is not entirely consistent on this point,
56
but this seems like the best way
to try to make him consistent.
One says that something is a mere idea if one cannot even approximate to it. 1. The
unconditioned of inherence (or of the aggregate). 2. That of dependence or of the
series. 3. That of the concurrence of all possibility in one and of one for all. (R 5553;
18:228)
56. The account of transcendental illusion in the introduction to the Dialectic and in
the chapter on Ideas is not altogether consistent with Kants remarks in the Antinomies.
For in the two former places he maintains that all ideas give rise to regressive series of
conditions. See especially A 3078/B 364 and A 336/B 393. It seems likely that in the B-
edition Kant simply failed to correct fully the parts of the text he did not wholly redraft.
488
Kants First Paralogism
Getting clearer about these subtle architectonic considerations
also helps with our second problem. This, recall, is the worry that in
the frst Critique Kant sometimes characterizes the illusion underlying the
paralogisms interms that make no reference to the apparent truthof D
or even to the apparent truth of an instance of D. The following remark
is typical of those moments:
Nothing is more natural and seductive than the illusion of taking the
unity in the synthesis of thoughts for a perceived unity in the subject of
these thoughts. (A 402)
What is in question here is an illusion prompting us to mistake one kind
of unity for another. This characterization of the illusion corresponds
closely to the one given in an important refection, from which the fol-
lowing two partial quotations are drawn:
The paralogism of pure reason is properly a transcendental subreption
[Subreption],
57
where our judgment about objects and the unity of con-
sciousness in them is held to be a perception of the unity of the subject.
(R 5553; 18:223)
The frst illusion [that is, the one operative in the Paralogisms, as opposed
to in the Antinomies or the Ideal of Pure Reason] is that where the unity
of apperception, which is subjective, is taken for the unity of the subject
as a thing. (R 5553; 18:224)
58
These remarks suggest, frst, that the unity in the synthesis of thoughts
mentioned at A 402 is the unity of consciousness (the unity of appercep-
tion), and, second, that the illusion prompting the paralogisms consists
57. By a subreption Kant means an instance of the fallacy vitium subreptionis (the
vice of theft or pilfering). One commits this fallacy when one mistakes a thing of one
kind for something of quite another kind. Kants examples of this fallacy are often cases in
which something subjective is mistaken for something objective, for example, a judgment
of perception is mistaken for a judgment of experience (24:767), or a conceptthe
subjective of thinkingis mistaken for a thingthe objective of what is thought. Real
Progress, 20:349. Kant also describes as a vitium subreptionis the fallacy from which arises
the mixing of concepts of experience and of reason (Blomberg Logic, sec. 255; 24:254
55), and here, by a concept of reason, he means a pure concept of the understanding.
Thus one commits this fallacy when, for example, one takes the concept of cause and
effect to be derived from experience. Ibid. The reverse errorthat of taking a sensitive
concept for a pure concept of the understandingalso counts as an instance of the
subreptive fallacy. See Inaugural Dissertation, 2:412. In the Dissertation, however, Kant is
explicit that he means to be using the term only by analogy with its accepted meaning. All
of this suggests that a transcendental subreption is the error of taking a thing of one kind
for a thing of another kind, where this error is grounded in transcendental illusion.
58. Erich Adickes dates this fragment to one of the periods 177879 and 178083.
489
I A N P R O O P S
in the appearance that this unity is the unity of the subject as a thing
(R 5553; 18:224).
What does the unity of the subject as a thing consist in? I would
suggest that, in the case of the frst paralogism, it is the unity apparently
imposed on experience in virtue of the subjects apparently instantiating
the category of inherence and subsistence. The notion of the unity of
the subject as a thing must, however, be interpreted differently in the dif-
ferent paralogisms. That much is clear from Kants table at A 404, which
sums up the ways in which each of the special categories
59
namely, the
categories of subsistence, reality, unity (as opposed to plurality), and exis-
tence (A 403)is supposed to secure its own kind of unconditioned
unity.
60
The table makes plain that each of these categories is supposed
to be paired with one of the attributes the rational psychologist ascribes
to the self. Thus, while the category of subsistence (and inherence) is
deployed in dialectical inferences to the substantiality of the self, that
of reality is deployed in dialectical inferences to its simplicity. ( Just why
this category goes with simplicity is not immediately apparent, as Kant
concedes in a footnote [A 404].) For its part, the category of unity (as
opposed to plurality) is deployed in inferences to the selfs identity over
time, while the category of existence is deployed in inferences to the selfs
independence from other things.
This style of diagnosis survives into the B-edition:
Rational psychology has its origin in a mere misunderstanding. The unity
of consciousness, which grounds the categories, is here taken for an intu-
ition of the subject as an object, and the category of substance is applied
to it. But this unity is only the unity of thinking, through which no object
is given; and thus the category of substance, which always presupposes a
given intuition, cannot be applied to it, and hence this subject cannot be
cognized at all. (B 42122)
How does this story about our tendency to confuse one kind of unity
with another connect up with the account of illusion given in terms of
59. Kant identifes these four categories as special because each of them is supposed
to ground the unity of the other categories in its class. A 403.
60. Rolf-Peter Horstmann sees Kant as countenancing several senses of unity: the
relational, qualitative, quantitative, and modal, corresponding, respectively, to the sub-
stantiality, simplicity, numerical identity, and plain existence of the soul. Horstmann
1993, 417. This is certainly along the right lines. However, because I fnd it doubtful that
Kant would have supposed the German word unity to be fourfold ambiguous, I prefer
to think of these as four distinct species of unityfour ways of being an unconditioned
condition.
490
Kants First Paralogism
the appearance that D, or an instance of it, is true? The connection is
not obvious, but a hint may be gleaned by considering two remarks from
the A-edition paralogisms chapter. Reason, Kant tells us, represents
[the categories of subsistence, reality, and so forth] as conditions of the
possibility of a thinking being which are themselves unconditioned (A
403). He also claims that the self is represented in transcendental illu-
sion as enjoying the unconditioned unity of relation, and he glosses
this as meaning not as inhering but rather subsisting (A 404). Taken
together, these remarks suggest that what it is for somethings to seem
to have categorially imposed unconditioned unity is for it to seem to
be unconditioned in some respect. The relevant respect will depend on
which paralogism is under consideration. In the second paralogism, for
example, the imagined unconditioned unity of the self amounts to its
imagined simplicity. In this case, transcendental illusion inclines us to
confuse the lack of complexity in the representation I with a lack of
complexity in a supposed objective self (A 381; compare A 340/B 398; A
355).
In general, then, reason is supposed to generate the illusion that
the self instantiates each of the special categories (subsistence, reality,
unity [as opposed to plurality], and existence) in virtue of its being an
unconditioned condition of the relevant kind (for example, an unbear-
able bearer of properties, a partless part, and so forth). Owing to tran-
scendental illusion, we are inclined to think of the self not just as unifying
its representations but, further, as unifying them in virtue of standing
to them as their unconditioned condition. This idea perhaps makes the
most sense inconnectionwiththe unconditioned unity of relation since
we can see why Kant might have taken transcendental illusion to tempt
us to view the self as an unbearable bearer of the thoughts that are its
predicates (A 346/B 404). In this case transcendental illusion tempts
us to confuse the unity of consciousness with the unconditioned unity
of relation, and so to suppose that the I of apperception represents
something that is indeed an unbearable bearer of properties, or
substance
1
. Thus the soul cognizes
61
. . . itself, not as inhering but rather
as subsisting (A 4034).
Because the idea of the self as an imagined unconditioned con-
dition remains even after the notion of its being the imagined uncon-
ditioned condition in a series is downplayed, it is possible to see Kants
various remarks about mistaking one kind of unity for another as indeed
61. This is, of course, a nonfactive use of erkennen.
491
I A N P R O O P S
in keeping with the spirit (if not, admittedly, the letter) of our origi-
nal account. This is possible because the illusion operative in the frst
paralogism can now be thought of simply as the illusion that there is
an unconditioned condition for any mental predicate in a particular
consciousnessthat is, a noninhering subject in which they all inhere.
Kant supposes that, when we are in the grip of this illusion, we will be
inclined to mistake the subjective unity of our thoughtsthe unity of
apperceptionfor the objective unity of a self. And this imagined objec-
tive unity will in this case just boil down to the selfs apparent property
of failing to be capable of inhering in another thing. However, because
principle D makes an essential reference to a series of conditions, the
present account is compatible only with a modifed version of our origi-
nal account. We must now say that P gives rise to the illusion not that D
is true but that another principle, D

, is:
D

: If the conditioned is given, then so is its unconditioned condi-
tion.
Kant recommends the Prolegomenas account of the paralogisms as supe-
rior to that of the frst edition of the frst Critique (4:381); but it is clear
fromhis remarks inthe Antinomies that he also entertained doubts about
the idea of characterizing transcendental illusion specifcally as the illu-
sion that a regressive series of conditions has an unconditioned last mem-
ber. Why should he have entertained these doubts?
Again, Kant does not say; but one possible reason is that such
a conception would invite the question: Why isnt there an antinomy of
the self ? This question would arise because, if the relation of inherence
were to generate a regressive series of conditions, the situation would
be structurally parallel to the case of an antinomy: we would be able to
seek the unconditioned either in a fnal member of the regress or , sup-
posing the regress unending, in an infnite totality of members subse-
quent in the regress to a given conditioned member. Kant in fact indi-
cates his awareness of these two options in the parenthetical remark in
the passage fromthe Prolegomena. He says, Pure reason demands that for
each predicate of a thing we should seek its own subject, but that for this
subject, which is in turn necessarily only a predicate, we should seek its
subject again, and so forth to infnity (or as far as we get) (Prolegomena,
sec. 46; 4:333, emphasis added). The availability of these two options,
each equally recommended by reason, would generate what Kant calls a
two-sided illusion, and so an antinomy. And yet in the frst Critique he
is keen to maintain that the transcendental paralogism effects a merely
492
Kants First Paralogism
one-sided illusion regarding the idea of the subject of our thought, and
for the opposite assertion there is not the least plausibility forthcoming
from concepts of reason (A 406/B 433). Kant is wedded to this view for
architectonic reasons. It is important to him to maintain that there are
only four antinomies, one for each heading of the table of categories (A
415/B 442; compare A 462/B 490). Kant sees this as important because
he wishes to claim that his critical project proceeds systematically. He
takes the neat division of antinomiesinto two mathematical antinomies
and two dynamical onesas a sign that he has succeeded in identifying
all the instances of dialectical inference associated with the hypothetical
syllogism.
Kants actual view in the Paralogisms avoids the appearance of a
two-sided illusion by dispensing with the picture of a regressive series;
but when fully thought through, this change demands a corresponding
change in the general account of transcendental illusion. Principle D

rather than D must now be adopted as the most general statement of the
content of transcendental illusion.
Conclusion
I have argued that Kants diagnosis of the rational metaphysicians dog-
matic belief in the substantiality
1
of the self in the B-edition paralogism
has a hitherto unappreciated persuasiveness. Through his account of
transcendental illusion, Kant has explained why sophisticated philoso-
phers might well have been attracted to this position. And he has, in addi-
tion, offered a suggestive diagnosis of the fallacy that constitutes the frst
paralogism. This diagnosis has two parts: the frst turns on the (correct)
idea that certain grammatical (or quasi-grammatical) considerations do
not, after all, function as a guide to conceivability. That, I take it, amounts
to an insight of lasting value. The second turns on the idea that we are apt
to confuse the unity of our perceptions and thoughts with the perception
of a unity. That idea can be seen as starting fromHumes observation that
we never do catch ourselves without a perception or idea. But, it goes
beyond this in offering a novel, non-Humean diagnosis of why we might
nonetheless imagine that we do so.
62
More questionable, I believe, is Kants attempt to portray transcen-
dental illusion, and the attendant doctrine of the transcendental ideas,
62. Humes own account of why we imagine ourselves to perceive a numerically iden-
tical, enduring, simple self is offered in book 1, part 4, section 6 of the Treatise. See Hume
2000, 16471.
493
I A N P R O O P S
as an exhaustive account of the origins of speculative metaphysics. Even
if we understand that discipline as narrowly as Kant does, such a claim
can seem hard to defend. Just confning ourselves to the topic of the
self, we might wonder: How can Kant be taken to have diagnosed, for
instance, Spinozas belief that the fnite mind is merely a mode of the
infnite mind of God? Wasnt Spinoza led by purely rational considera-
tions to the doctrinea dogma by Kants lightsthat the human soul
is not a substance
1
? Or, again: How can he be taken to have diagnosed the
Cartesian doctrine that the essence of the self is to think? Kant believes
that the self can be known to think but not that it can be known to be
incapable of existing without thought. One searches the paralogisms in
vain for any diagnosis of how such doctrines might arise.
It seems likely that the pressures of Kants architectonic play a role
here. In respecting the constraints laid down by that architectonic, Kant
simply lacks the fexibility to offer diagnoses of the full range of ratio-
nalist doctrines. Nonetheless, I think it fair to say that Kant has at least
provided an elegant and somewhat plausible account of how in princi-
ple philosophers might arrive at the belief that the self is a substance
1
.
Whether that account accurately describes the actual reasoning of indi-
vidual proponents of that view is, of course, a further question; but one,
unfortunately, that lies beyond the scope of this essay.
References
Ameriks, Karl. 2000 [1982]. Kants Theory of Mind: An Analysis of the Paralogisms
of Pure Reason. 2nd ed. Oxford: Oxford University Press.
Aquinas, Thomas. 1999. Commentary on Aristotles Physics, trans. R. J. Blackwell,
R. J. Spath, and W. Edmund Thirlkel. Rev. ed. South Bend, IN: Dumb Ox
Books.
Aristotle. 1984. The Complete Works of Aristotle: The Revised Oxford Translation, ed.
Jonathan Barnes. Vol. 1. Princeton: Princeton University Press.
Baumgarten, Alexander. 1902. Metaphysica. Reprinted in Immanuel Kant,
Gesammelte Schriften, vol. 15, part 1 (554) and vol. 17 (5226). Berlin:
K oniglich-Preussischen Akademie der Wissenschaften zu Berlin, 1902.
Orig. pub. in Halle, 1739, and Frankfurt, 1757. English translation of
selected passages in Eric Watkins, Kants Critique of Pure Reason: Background
Source Materials. Cambridge: Cambridge University Press, 2009.
Bennett, Jonathan. 1966. Kants Analytic. Cambridge: Cambridge University
Press.
.1974. Kants Dialectic. Cambridge: Cambridge University Press.
Boehm, Omri. 2009. Kants Critique of Spinoza. PhD diss., Yale University.
494
Kants First Paralogism
Chignell, Andrew. 2007. Belief in Kant. Philosophical Review 116: 32360.
Erdmann, Benno. 1881. Nachtr age zu Kants Kritik der reinen Vernunft. Kiel: Lipsius
and Tischer.
Grier, Michelle. 1993. Illusion and Fallacy in Kants First Paralogism. Kant
Studien 83: 25782.
. 2001. Kants Doctrine of Transcendental Illusion. Cambridge: Cambridge
University Press.
Guyer, Paul, ed. 2005. Notes and Fragments. The Cambridge Edition of the
Works of Immanuel Kant. Trans. Curtis Bowman, Paul Guyer, and Frederick
Rauscher. Cambridge: Cambridge University Press.
Guyer, Paul, and Allen Wood, eds. and trans. 1998. Critique of Pure Reason.
The Cambridge Edition of the Works of Immanuel Kant. Cambridge: Cam-
bridge University Press.
Horstmann, Rolf-Peter. 1993. Kants Paralogismen. Kant-Studien 84: 40825.
Hume, David. 2000. A Treatise of Human Nature, ed. D. F. Norton and M. J.
Norton. Oxford: Oxford University Press.
Kant, Immanuel. 1902. Gesammelte Schriften. Berlin: K oniglich-Preussischen
Akademie der Wissenschaften zu Berlin (now de Gruyter).
Kemp Smith, Norman, trans. 1933. Kants Critique of Pure Reason. 2nd ed. Lon-
don: Macmillan. Repr., Boston: Bedford Books, 1969.
Kitcher, Patricia. 1982. Kants Paralogisms. Philosophical Review 91: 51547.
Longuenesse, B eatrice. 2008. Kants I think versus Descartes I am a thing
that thinks. In Kant and the Early Moderns, ed. Daniel Garber and B eatrice
Longuenesse, 931. Princeton: Princeton University Press.
Pinder, Tillmann. 1987. Zu Kants Logik-Vorlesung um 1780, anl asslich einer
neu aufgefundenen Nachschrift. Kant-Forschungen 1: 79114.
Spinoza. 1992. The Ethics; Treatise on the Emendation of the Intellect; Selected Letters,
ed. S. Feldman, trans. S. Shirley. Indianapolis: Hackett.
van Cleve, James. 1999. Problems from Kant. Oxford: Oxford University Press.
Wilson, Margaret. 1974. Leibniz and Materialism. Canadian Journal of Philoso-
phy 3: 495513.
Wolff, Christian. 1747. Vern unftige Gedancken von Gott, der Welt und der Seele des
Menschen, auch allen Dingen uberhaupt [Rational Thoughts onGod, the World
and the Soul of Human Beings, also All Things in General]. Halle: Renger.
Orig. pub. 1720.
Young, J. Michael, ed. and trans. 1992. Lectures on Logic. The Cambridge Edition
of the Works of Immanuel Kant. Cambridge: Cambridge University Press.
495

You might also like