<<

ńňđ. 4
(âńĺăî 10)

ŃÎÄĹĐĆŔÍČĹ

>>

dimensions are always present for Kant.
TRANSCENDENTAL ANALYTIC
102

Here it will be useful to recall the problem laid out in the letter to Herz
mentioned in the first section of this chapter. Mathematical concepts
present their own objects by directing the synthesis of an a priori (spatial)
manifold according to rules provided by the relevant concept (e.g. a line,
a triangle, a circle). But we cannot do that in metaphysics, because there
the objects of our concepts are not just constructed in pure intuition.
They are supposed to be independently existing things, so that in this
case we just do not see how a priori concepts might relate to objects.31
Well, here (in section three of the Leitfaden) Kant is telling us that a
function of the understanding, the function of judging, is not arbitrarily
producing (constructing) representations of objects, as in geometry or
even in arithmetic, but at least unifying according to rules the presented
manifold of intuition, so that it can be analyzed into (empirical) concepts
and thought about in judgments.
Thus he writes:

Synthesis in general, as we shall subsequently see, is the mere effect of
imagination, a blind though indispensable function of the soul, without
which we would have no cognition at all, but of which we are seldom even
conscious. Yet to bring this synthesis to concepts is a function that pertains to the
understanding [my emphasis] and by means of which it first provides us
with cognition in the proper sense. (A78/B103)

What might it mean, to ‘‘bring synthesis to concepts’’? I suggest the
following. What is given to us in sensibility is given in a dispersed way –
spread out in space and in time, where similar things do not present
themselves to us at the same time but rather, need to be recalled in order
to be compared. Moreover, the variety and variability of what does
present itself is such that which pattern of regularity should be picked
out might be anybody’s guess. Even the way we synthesize or bind
together the manifold might itself be quite random, obeying here
some rule of habitual association, there some emotional connection,
and so on. So ordering the synthesis itself under systematic rules
so that the components of intuition can be thought under common
concepts in a regular fashion is the work of the understanding. The
understanding thus ‘‘brings synthesis to concepts.’’ It makes it the case
that synthesis does give rise to, opens the way for, conceptualization.



31
See above, p. 87. Cf. Correspondence, AAx , p. 131.
METAPHYSICAL DEDUCTION OF THE CATEGORIES 103

The analogy with the mathematical case is only partly helpful here.
Kant writes:

Now pure synthesis, universally represented, yields the pure concept of
the understanding. By this synthesis, however, I understand that which
rests on a ground of synthetic unity a priori: thus our counting (as is
especially noticeable in the case of larger numbers) is a synthesis in
accordance with concepts, since it takes place in accordance with a
common ground of unity (e.g. the decimal). Under this concept, there-
fore, the synthesis of the manifold becomes necessary. (A78/B104)

In counting, we add unit to unit, and then units of higher order
(a decade, a hundred, a thousand, and so on) that allow us to synthesize
(enumerate) larger and larger collections (of items, of portions of
a line . . . ). The idea is that similarly, in ordering empirical manifolds,
we make use of grounds of unity of these manifolds (say: whenever event
of type A occurs, then event of type B also occurs), which we think under
concepts or ‘‘represent universally’’ (in the case at hand, under the
concept of cause). We thus form chains of connections between these
manifolds, in an effort to unify them in one space and one time, in the
context of one and the same totality of experience. But of course,
whereas it is always possible to enumerate a collection of things or
parts of things once one has arbitrarily given oneself a unit for counting
or measuring, in contrast, actually finding repeated occurrences of
similar events depends on what experience presents to us. Because of
this difference, Kant distinguishes the former kind of synthesis, which he
calls ‘‘mathematical’’ synthesis, from the latter, which he calls ‘‘dynami-
cal,’’ and he accordingly distinguishes the corresponding categories by
dividing them along the same line (see B110; A178–9/B221–2).
Nevertheless, in the latter case just as in the former, a ‘‘ground of
unity’’ that has its source in the understanding is at work in our synthe-
sizing (combining, relating) the objects of our experience or their spatio-
temporal parts. This ground of unity, says Kant, is a pure concept of the
understanding.
This reasoning leads to the core statement of all three sections of the
Leitfaden chapter:

The same function that gives unity to the different representations in
a judgment also gives unity to the mere synthesis of different representa-
tions in an intuition, which, expressed universally, is called the pure
concept of understanding. The same understanding, therefore, and
TRANSCENDENTAL ANALYTIC
104

indeed by means of the very same actions through which it brings the
logical form of a judgment into concepts by means of the analytical unity,
also brings a transcendental content into its representations by means of
the synthetic unity of the manifold in intuition in general, on account of
which they are called pure concepts of the understanding that pertain
a priori to objects; this can never be accomplished by general logic.
(A79/B104–5)

I indicated above how the introduction of the term function at the
beginning of section one already foreshadowed the argument of section
three: the very same ‘‘unity of the act’’ that accounts for the unity of
concepts of judgments also accounts for there being just those forms of
unity in our intuitions that make them liable to being reflected under
concepts in judgment. The concepts that reflect those forms of unity in
intuition are the categories. But they do not just reflect those forms of
intuitive unity. As the mathematical analogue made clear (cf. A78/B104
cited above), they originally guide them. So for instance, as we just saw,
the concept of magnitude is that concept that guides the operation of
finding (homogeneous) units (say, points, or apples) or as the case may
be, units of measurement (say, a meter) and adding them to one another
in enumerating a collection or in measuring a line. The end result of this
operation is the determination of a magnitude, whether discrete (the
number of a collection) or continuous (the measurement of a line) as
when we say that the number of pears on the table is seven or the
measurement of the line is 4 meters. Here we reflect the successive
synthesis of homogeneous units under the concept of a determinate
magnitude (7 units, 4 meters). Similarly, the concept of cause (the con-
cept of some event’s being such as to be adequately or ‘‘in itself’’ reflected
under the antecedent of a hypothetical judgment with respect to another
event, adequately or ‘‘in itself’’ reflected under the consequent) guides
the search for some event that might always precede another in the
temporal order of experience. Once such a constant correlation is
found, we say that event of type A is the cause of event of type B. In
other words, the sequence is now reflected under the concept of a
determinate causal connection.32


32
In the chapter on the Schematism of the Pure Concepts of the Understanding, Kant
maintains that the schema of the concept of cause is ‘‘the real upon which, when it is
posited, something else always follows’’ (A144/B183). This means that it is by apprehend-
ing the regular repetition of a sequence of events or states of affairs (‘‘the real upon which,
METAPHYSICAL DEDUCTION OF THE CATEGORIES 105

The two aspects in our use of categories are explicitly mentioned in x10.
Kant says, on the one hand, that categories ‘‘give unity to [the] pure synth-
esis’’ (A79/B104). He says, on the other hand, that the pure concepts of the
understanding are ‘‘the pure synthesis generally represented ’’ (A78/B104; see
also A79/B105 quoted earlier, where both aspects are present in one and
the same sentence: ‘‘the same function . . . gives unity which expressed gene-
rally, is the pure concept of the understanding’’). These two points are fully
explained only in book two of the Transcendental Analytic, ‘‘The Analytic
of Principles.’’ There Kant explains that categories, insofar as they deter-
mine rules for synthesis of sensible intuitions, have schemata (ch. 1 of book
two, A137/B176). Being able to pick out instances of such schemata allows
us to subsume our intuitions under the categories (ch. 2 of book two, A148/
B187–A235/B287). Only in those chapters does Kant give a detailed
account of the way in which each category both determines and reflects a
specific rule (a schema) for the synthesis of intuitions.
As far as the metaphysical deduction is concerned, Kant is content with
making the general case that:
In such a way there arise [entspringen] exactly as many pure concepts of
the understanding which apply to objects of intuition a priori, as there
were logical functions of all possible judgments in the previous table: for
the understanding is completely exhausted, and its capacity entirely
measured by these functions. (A80/B106)

Kant does not mean that every time we make use of a particular logical
function/form of judgment, we thereby make use of the corresponding
category. True, absent a sensible manifold to synthesize, all that remains
of the categories are logical functions of judgment. But the logical func-
tions of judgment are not, on their own as it were, categories. They
become categories (categories ‘‘arise,’’ entspringen, as Kant says in the text

whenever posited, something else follows’’) that we recognize in experience the presence
of a causal connection. But conversely, we look for such constant conjunctions because we
do have a concept of cause as the concept of something that might be thought under the
antecedent of a hypothetical judgment, with respect to something else that might be
thought under the consequent. Of course Kant’s point is also that we can always be
mistaken about what we so identify. Some repeated sequence is warranted as a true causal
connection only if it can be thought under a causal law, and this involves the application of
mathematical constructions that allow us to anticipate the continuous succession and
correlation of events in space and in time. However, here I am anticipating developments
of Kant’s argument that go way beyond the metaphysical deduction properly speaking.
See my ‘‘Kant on causality: what was he trying to prove?’’ in Christia Mercer and Eileen
O’Neill (eds.), Early Modern Philosophy: Mind, Matter and Metaphysics (Oxford: Oxford
University Press, 2005); reprinted as ch. 6 in this volume.
TRANSCENDENTAL ANALYTIC
106

just cited) only when the understanding’s capacity to judge is applied to
sensible manifolds, thus synthesizing them (combining them in intui-
tion) for analysis (into concepts) for synthesis (of concepts in judgments).
And even then, there remains a difference between the category’s guid-
ing the synthesis of manifolds, and the manifolds’ being correctly
subsumed under the relevant category. For instance, it may be the case
that the understanding’s effort to identify what might fall under the
antecedent and what might fall under the consequent of a hypothetical
judgment, leads it to recognize the fact that whenever the sun shines on
the stone, the stone gets warm. This by itself does not warrant the claim
that there is an objective connection (a causal connection) between the
light of the sun and the warmth of the stone. Only some representation
of the overall unity of connections of events in the world can give us at
least a provisional, revisable warrant that this connection is the right one
to draw.33
Kant is not yet explaining how his metaphysical deduction of the
categories might put us on the way to resolving the problem left open
after the 1770 Inaugural Dissertation: how do concepts that have their
source in the understanding apply to objects that are given? All we have
here is an exposition as a system ‘‘from a common principle, namely the
capacity to judge’’ (A80–1/B106) of the table of the categories, and an
explanation of the role they perform in synthesizing manifolds so that
the latter can be reflected under concepts combined in judgments. To
respond to the problem he set himself, Kant will need to argue that those
combining activities are necessary conditions for any object at all to
become an object of cognition for us. And as I suggested earlier, only
the later argument will provide a full justification of the table of logical
forms itself: it is a table making manifest just those functions of judging
that are necessary for any empirical concept at all to be formed by us,
and thus for any empirical object to be recognized under a concept.
This confirms again that the ‘‘leading thread’’ from logical forms to
categories is precisely no more (but no less) than a ‘‘leading thread.’’ Its
actual relevance will be proved only when the argument of the
Transcendental Deduction is expounded and in turn, opens the way to
the Schematism and System of Principles.




33
On this example, see Prolegomena, AAiv, pp. 312–13. See also the related discussion above,
ch. 2, pp. 58–62.
METAPHYSICAL DEDUCTION OF THE CATEGORIES 107


The impact of Kant’s metaphysical deduction of the categories
The history of Kant’s metaphysical deduction of the categories is not a
happy one. Kant’s idea that a table of logical functions of judgments
might serve as a leading thread for a table of the categories was very early
on an object of suspicion, on three main grounds. First, Kant’s careless
statement that he ‘‘found in the labors of the logicians,’’ namely in the
logic textbooks of the time, everything he needed to establish his table of
the logical forms of judgment raises the obvious objection that the latter
is itself lacking in systematic justification.34 This in turn casts doubt on
Kant’s claim that unlike Aristotle’s ‘‘rhapsodic’’ list (A81–2/B106–7), his
table of the categories is systematically justified. Second, even if one does
endorse Kant’s table of the logical forms of judgment, this does not
necessarily make it an adequate warrant for his table of the categories.
And finally, once the Aristotelian model of subject–predicate logic was
challenged by post-Fregean truth-functional, extensional logic, it
seemed that the whole Kantian enterprise of establishing a table of
categories according to the leading thread of forms pertaining to the
old logic seemed definitively doomed.
An early and vigorous expression of the first charge mentioned above
was Hegel’s. In the Science of Logic, Hegel writes:
Kantian philosophy . . . borrows the categories, as so-called root notions
for transcendental logic, from subjective logic in which they were
adopted empirically. Since it admits this fact, it is hard to see why
transcendental logic chooses to borrow from such a science instead of
directly resorting to experience.35

Note, however, that it is not Kant’s table of logical forms per se that
Hegel objects to. Rather, it is the way the table is justified (or rather, not
justified) and the random, empirical way in which the categories them-
selves are therefore listed. Nevertheless, in the first section of his
Subjective Logic, Hegel too expounds four titles and for each title,
three divisions of judgment that exactly map the titles and divisions of

34
Cf. Prolegomena, AAiv, pp. 323–4.
35
G. W. F. Hegel, Wissenschaft der Logik, ii: Die subjective Logik, in Gesammelte Werke, Deutsche
Forschungsgemeinschaft, ed. Rhein-Westfal. Akad. d.Wiss. (Hamburg: F. Meiner,
¨
1968–), vol. xii, pp. 253–4; Science of Logic, trans. A. V. Miller (Atlantic Highlands, NJ:
Humanities Press International, 1989), p. 613. What Hegel means here by ‘‘subjective
logic’’ is what Kant called ‘‘pure general logic,’’ namely the logic of concepts, judgments,
and syllogistic inferences. But unlike Kant’s ‘‘pure general logic,’’ Hegel’s subjective logic
is definitely not ‘‘merely formal.’’ More on this shortly.
TRANSCENDENTAL ANALYTIC
108

Kant’s table, although Hegel starts with the title of quality rather than
quantity. Moreover, the names of each title are changed, although the
names of the divisions remain the same. Kant’s title of ‘‘quality’’ becomes
‘‘judgment of determinate-being’’ (Urteil des Daseins), with the three
divisions of positive, negative, and infinite judgment. ‘‘Quantity’’
becomes ‘‘judgment of reflection’’ with the three titles of singular, parti-
cular, and universal. ‘‘Relation’’ becomes ‘‘judgment of necessity’’ (sic!)
with the three titles of categorical, hypothetical, and disjunctive. And
finally ‘‘modality’’ becomes ‘‘judgment of the concept’’ with the three
divisions of assertoric, problematic, and apodeictic.36 Of course, the
change in nomenclature signals fundamental differences between
Hegel’s and Kant’s understanding of the four titles and their twelve
divisions. The most important of those differences is that for Hegel the
four titles and three divisions within each title do not list mere forms of
judgment, but forms with a content, where content and form are
mutually determining. So for instance, the content of ‘‘judgments of
determinate-being’’ (affirmative, negative, infinite) is the immediate,
sensory qualities of things as they present themselves in experience.
The content of ‘‘judgments of reflection’’ (singular, particular, universal)
is what Hegel calls ‘‘determinations of reflection,’’ namely general repre-
sentations, or representations of common properties as they emerge for
an understanding that compares, reflects, abstracts. The content of
‘‘judgments of necessity’’ (categorical, hypothetical, disjunctive) is the
relation between essential and accidental determinations of things.
And finally the content of ‘‘judgments of the concept’’ (assertoric,
problematic, apodeictic) is the normative evaluation of the adequacy of
a thing to what it ought to be, or its concept. So certainly Hegel’s
interpretation of each title radically transforms its Kantian ancestor.
Nevertheless, the fact that despite his criticism of Kant’s empirical
derivation, Hegel maintains the structure of Kant’s divisions, indicates
that his intention is not to criticize the classifications themselves, but rather
to denounce the cavalier way in which Kant asks us to accept them as well
as Kant’s shallow separation between form and content of judgment.37
Nor is Hegel’s intention to challenge the relation between categories
and functions of judgment. In the Science of Logic, categories of quantity
and quality are expounded in part one (Being) of book one (The

36
See Die subjective Logik, pp. 59–90; trans. pp. 623–63.
37
On this point see my ‘‘Hegel, Lecteur de Kant sur le jugement,’’ in Philosophie, no. 36
(1992), pp. 62–7.
METAPHYSICAL DEDUCTION OF THE CATEGORIES 109

Objective Logic); those of relation and modality are expounded in part
two (The Doctrine of Essence) of book one. Logical forms of judgment
and syllogistic inference are expounded in section one of book two (The
Subjective Logic or the Doctrine of the Concept). If we accept, as I
suggest we should, that book two expounds the activities of thinking
that have governed the revelation of the categorical features expounded
in parts one and two of book one, then Hegel’s view of the relation
between categories and forms of judgment is similar to Kant’s at least
in one respect: there is a fundamental relation (in need of clarification)
between the structural features of the acts of judging and the structural
features of objects. The difference between Hegel’s view and Kant’s view
is that Hegel takes this relation to be a fact about being itself, and the
structures thus revealed to be those of being itself, whereas Kant takes
the relation between judging and structures of being to be a fact about
the way human beings relate to being, and the structures thus revealed
to be those of being as it appears to human beings.
Hegel’s grandiose reinterpretation of Kant’s titles of judgments did
not have any immediate posterity, and his speculative philosophy was
soon superseded by the rise of naturalism in nineteenth-century philo-
sophy.38 When Hermann Cohen, reacting against both the excesses of
German Idealism and the rampant naturalism of his time, undertook to
revive the Kantian transcendental project, he declared that his goal was
to ‘‘ground anew the Kantian theory of the a priori’’ (‘‘die Kantische
Aprioritatslehre erneut zu begrunden’’).39 By this he meant that, against
¨ ¨
the vagaries of Kant’s German Idealist successors, he intended to lay out
what truly grounds Kant’s theory of the categories and a priori princi-
ples. According to Cohen, Kant’s purpose in the Critique of Pure Reason is
to expound the presuppositions of the mathematical science of nature
founded by Galileo and Newton. The leading thread for Kant’s pure
concepts of the understanding or categories (expounded in book one of
the Transcendental Analytic) is really Kant’s discovery of the principles
of pure understanding (expounded in book two), and the leading thread
for the latter are Newton’s principles of motion in the Principia
Mathematica Philosophiae Naturalis. Thus the true order of discovery of
the Transcendental Analytic leads from the Principles of Pure
Understanding (book two), to the Categories (book one). This does not

38
On this point, see Hans D. Sluga, Gottlob Frege (London: Routledge and Kegan Paul, 1980),
pp. 8–35.
39
Cohen, Kants Theorie der Erfahrung, p. ix.
TRANSCENDENTAL ANALYTIC
110

make the logical forms of judgment irrelevant, in Cohen’s eyes. For the
latter formulate the most universal patterns or models of thought
derived from the unity of consciousness, which for Cohen is nothing
other than the epistemic unity of all principles of experience, where
experience means scientific knowledge of nature expounded in
Newtonian science. So it is quite legitimate to assert that the categories
depend on these universal patterns. But the systematic unity of the
categories and of the logical forms can be discovered only by paying
attention to the unity of the principles of the possibility of experience, i.e.
of the Newtonian science of nature.40
Cohen follows up on his interpretative program by showing how
Kant’s systematic correlation between logical forms of judgment and
categories can be understood in the light of the distinction he offers in
the Prolegomena between judgments of perception and judgments of
experience. Cohen then proceeds to explain and justify Kant’s selection
of logical forms by relating each of them to the corresponding category
and to its role in the constitution of experience. In other words, he
implements the very reversal in the order of exposition that he argues
is faithful to Kant’s true method of discovery: moving from the a priori
principles that may ground judgments of experience, to the categories
present in the formulation of these principles, to the logical forms of
judgment.41
Cohen’s achievement is impressive. But it is all too easy to object that
his reducing Kant’s unity of consciousness to the unity of the principles
of scientific knowledge, and his reducing Kant’s project to uncovering
the a priori principles of Newtonian science, amount to a very biased
reading of Kant’s Critique of Pure Reason. In fairness to Cohen, his
interpretation of Kant’s critical philosophy did not stop there. In Kants
¨ndung der Ethik,42 he considered Kant’s view of reason and its
Begru
role in morality. And this in turn led him to give greater consideration,
in the second and third editions of Kants Theorie der Erfahrung, to Kant’s
theory of the ideas of pure reason and to the bridge between knowledge
and morality.43 Nevertheless, as far as the metaphysical deduction of
the categories is concerned, his interpretation remained essentially
unchanged.

40
Cohen, Kants Theorie der Erfahrung, p. 229.
41
Ibid., pp. 245–8.
42
Hermann Cohen, Kants Begrundung der Ethik (Berlin, 1877; 2nd edn 1910).
¨
43
See Kants Theorie der Erfahrung, preface to the second edition, p. xiv.
METAPHYSICAL DEDUCTION OF THE CATEGORIES 111

That interpretation found its most vigorous challenge in Heidegger’s
reading of Kant’s first Critique. Heidegger urges that Kant did not intend
his Critique of Pure Reason primarily to clarify the conceptual presuppositions
of natural science. Rather, Kant’s goal was to question the nature and
possibility of metaphysics. According to Heidegger, this means laying out
the ontological knowledge (knowledge of being as such) that is presupposed
in all ontic knowledge (knowledge of particular entities). Kant’s doctrine of
the categories is precisely Kant’s ‘‘refoundation’’ of metaphysics, or his effort
to find for metaphysics the grounding that his predecessors had been
unable to find. This refoundation consists, according to Heidegger, in
elucidating the features of human existence in the context of which
human beings’ practical and cognitive access to being is made possible.
What does this have to do with Kant’s enterprise in the metaphysical
deduction of the categories? In the Phenomenological Interpretation of the
Critique of Pure Reason (a lecture course delivered at Marburg in 1927–8,
and first published in 1977) and in Kant and the Problem of Metaphysics
(first edition, 1929), Heidegger develops the following view. Kant’s
groundbreaking insight was to discover that the unity of our intuitions
of space and of time, and the unity of concepts in judgments, have one
and the same ‘‘common root’’: the synthesis of imagination in which
human beings develop a unified view of themselves and of other entities
as essentially temporal entities. Now, categories, according to
Heidegger, are the fundamental structural features of the unifying
synthesis of imagination which results in the unity of time (and space)
in intuition, on the one hand; and in the unity of discursive representa-
tions (concepts) in judgments, on the other hand. This being so, the
fundamental nature of the categories is expounded not in the metaphy-
sical deduction, which relates categories to logical forms of judgments,
but rather in the Transcendental Deduction and even more in
the chapter on the Schematism of the Pure Concepts of the
Understanding. For it is in these two chapters that the role of the
categories as structuring human imagination’s synthesizing (unifying)
of time is expounded and argued for. This does not mean that the
Metaphysical Deduction is a useless or irrelevant chapter of the
Critique. For if it is true that the unity of intuition and the unity of
judgments have one and the same source in the synthesis of imagination
according to the categories, then the logical forms of judgment do give a
clue to a corresponding list of the categories. But this should not lead to
the mistaken conclusion that the categories have their origin in logical
forms of judgment. Rather, logical forms of judgment give us a clue to
TRANSCENDENTAL ANALYTIC
112

those underlying forms or structures of unity because they are the sur-
face effect, as it were, of forms of unity that are also present in sensibility
(where they are manifest as the schemata of the categories) by virtue of
one and the same common root in imagination.44
Note that Heidegger agrees with Cohen at least in maintaining that
logical forms of judgment can provide a leading thread to a table of
categories just because forms of judgment and categories have one and
the same ground, the unity of consciousness. Their difference consists in
the fact that Cohen understands that unity as being the unity of thought
expressed in the principles of natural science. Heidegger understands it
as the unity of human existence projecting the structures of its own
temporality.
The readings of Kant’s metaphysical deduction we have considered so
far offer challenges only to Kant’s motivation and method in adopting a
table of logical forms of judgment as the leading thread to his table of
categories. What they do not challenge is the relevance of Kant’s
Aristotelian model of logic in developing the argument for his table of
the categories. A more radical challenge comes of course from the idea
that contrary to Kant’s claim, logic did not emerge in its completed and
perfected form from Aristotle’s mind (cf. Bviii). Here we have to make a
quick step back in time. For the initiator of modern logic, Gottlob Frege,
wrote his Begriffsschrift (1879) several decades before Heidegger wrote
Being and Time (1927). Unsurprisingly, by far the more threatening
challenge to Kant’s metaphysical deduction came from Frege’s
Begriffsschrift and its aftermath.
As we saw, Kant takes logic to be a ‘‘science of the rules of the under-
standing.’’ But Frege takes logic to be the science of objective relations of
implication between thoughts or what he calls ‘‘judgeable contents.’’45


44
See Martin Heidegger, Phanomenologische Interpretation der Kritik der reinen Vernunft, col-
¨
lected edn vol. xxv (Frankfurt-am-Main: Vittorio Klostermann, 1977), pp. 257–303;
Phenomenological Interpretation of the Critique of Pure Reason, trans. Parvis Emad and
Kenneth Maly (Bloomington and Indianapolis: Indiana University Press, 1995),
pp. 175–207. And Kant und das Problem der Metaphysik, collected edn vol. iii (Frankfurt-
am-Main: Vittorio Klostermann, 1991), pp. 51–69; Kant and the Problem of Metaphysics,
trans. Richard Taft (Bloomington: Indiana University Press, 1990), pp. 34–46.
45
Gottlob Frege, Begriffsschrift. Eine der arithmetischen nachgebildete Formelsprache des reinen
Denkens, in Begriffsschrift und andere Aufsatze (Hildesheim: Olms, 1964). Begriffsschrift, a
¨
formula language for pure thought, modeled upon that of arithmetic, in Frege and Godel: Two
¨
Fundamental Texts in Mathematical Logic, ed. Jean van Heijenhoort (Cambridge, Mass.:
Harvard University Press, 1970). Page references will be to the English edition. On the
distinction between judgment and judgeable content, see ibid., x2, p. 11:
METAPHYSICAL DEDUCTION OF THE CATEGORIES 113

Against the naturalism that tended to become prevalent in nineteenth-
century views of logic, Frege defends a radical distinction between the
subjective conditions of the act of thinking and its objective content.
Logic, according to him, is concerned with the latter, psychology with
the former. In spite of his declared intention not to mix general pure
(ÂĽ formal) logic with psychology, Kant, according to Frege, is confused
in maintaining that logic deals with the rules we (human beings) follow in
thinking, rather than with the laws that connect thoughts independently
of the way any particular thinker or group of thinkers actually think.46
According to Frege, Kant’s subservience to the traditional, Aristotelian
model of subject–predicate logic is grounded on that confusion. For the
subject–predicate model really takes its clue from the grammatical struc-
ture of sentences in ordinary language. And ordinary language is itself
governed by the subjective, psychological intentions and associations of
the speaker addressing a listener. But again, what matters to logic are the
structures of thought that are relevant to valid inference, nothing else.
Those structures, for Frege, include the logical constants of propositional
calculus (negation and the conditional), the analysis of propositions into
function-argument rather than subject–predicate, and quantification.47
In x4 of the Begriffsschrift, Frege examines ‘‘the meaning of distinctions
made with respect to judgments.’’ The distinctions in question are clearly
those of the Kantian table, which in Frege’s time have become classic.
Frege first notes that those distinctions apply to the ‘‘judgeable content’’
rather than to judgment itself.48 This being said, he retains as relevant to
logic the distinction between ‘‘universal’’ and ‘‘particular’’ judgeable

‘‘A judgment will always be expressed by means of the sign , which stands to the left of
the sign, or the combination of signs, indicating the content of the judgment. If we omit the
small vertical stroke at the left end of the horizontal one, the judgment will be transformed
into a mere combination of ideas [Vorstellungsverbindung], of which the writer does not state
whether he acknowledges it to be true or not.’’
Later Frege renounces the expression Vorstellungsverbindung as too psychological, and
talks instead of Gedanke to describe the judgeable content to the right of the judgment
stroke. See the 1910 footnote Frege appended to x2, p. 11, n. 6.
46
On the rise of nineteenth-century naturalism about logic, and Frege’s conception of logic
as a reaction against naturalism, see Sluga, Frege, especially ch. 1 and 2. In fairness to Kant,
it should be recalled that he does distinguish logic from psychology: he maintains that
contrary to the latter, the former is concerned not with the way we think, but with the way
we ought to think. But this distinction can have little weight for Frege, who wants to free
logic from any mentalistic connotation, whether normative or descriptive.
47
Strawson’s criticism of the redundancies of Kant’s table is clearly inspired from Frege’s.
See Strawson, Bounds of Sense, pp. 78–82.
48
It is worth noting that Frege reverses the Kantian terminology and calls ‘‘proposition’’ the
judgeable content and ‘‘judgment’’ the asserted content, whereas Kant reserved the term
TRANSCENDENTAL ANALYTIC
114

contents (Kant’s first two titles of quantity), but leaves out ‘‘singular.’’ He
retains ‘‘negation’’ (Kant’s second title of quality, negative judgment) and
thus the contrasting affirmation (which does not need any specific nota-
tion), but leaves out infinite judgments. He declares that the distinction
between categorical, hypothetical, and disjunctive judgments ‘‘seems to
me to have only grammatical significance.’’ Meanwhile he introduces his
own notation for conditionality in the next section, x5 of the Begriffsschrift
(more on this in a moment). Finally, he urges that the distinction
between assertoric and apodeictic modalities (which alone, he says,
characterize judgment rather than merely the judgeable content)
depends only on whether the judgment can be derived from a universal
judgment taken as a premise (which would make the judgment apodeic-
tic), or not (which would leave it as a mere assertion, or assertoric
judgment), so that this distinction ‘‘does not affect the conceptual
content.’’ Frege presumably means that the distinction between assertoric
and apodeictic judgments does not call for a particular notation in the
Begriffsschrift. As for a proposition ‘‘presented as possible,’’ Frege takes it to
be either a proposition whose negation follows from no known universal
law, or a proposition whose negation asserted universally is false. Although
this last characterization differs from Kant’s characterization of prob-
lematic judgments (as components in hypothetical or disjunctive judg-
ments), it remains that Frege’s view of modality is similar to Kant’s own
view, indeed seems inspired by it. For as we saw Kant thinks that modality
does not concern the content of any individual judgment, but only its
relation to the unity of thought in general. However, Kant does not think
that what we might call this ‘‘holistic’’ view of modality makes it irrelevant to
logic. This point would be worth pursuing, but we cannot do it here.
In short, according to Frege one need retain from the Kantian table
only the first two titles of quantity, the first two titles of quality, and the
second title of modality (assertion expressed by the judgment stroke). To
these he adds his own operator of conditionality, which one might think
has a superficial similarity to Kant’s hypothetical judgment. However,
Frege makes it clear they are actually quite different. He recognizes
explicitly, for instance, that his conditional is not the hypothetical judg-
ment of ordinary language, which he identifies with Kant’s hypothetical
judgment. And he states that the hypothetical judgment of ordinary
language (or Kant’s hypothetical judgment) expresses causality.49
‘‘proposition’’ to assertoric judgment: see above, n. 18; Begriffsschrift, x2, x4. These are
mere terminological differences, but they need to be kept in mind to avoid confusions.
49
Begriffsschrift, x5, p. 15.
METAPHYSICAL DEDUCTION OF THE CATEGORIES 115

However, his view on this point does not seem to be completely fixed, at
least in the Begriffsschrift, since elsewhere in this text he urges that the
causal connection is expressed by a universally quantified conditional.50
In any event, Kant would not accept any of those statements. For as we
saw, he would say that although the hypothetical judgment does express
a relation of Konsequenz between antecedent and consequent, this rela-
tion is not by itself sufficient to define a causal connection. As for the
universal quantification of a conditional, it would even less be sufficient
to express a causal connection, precisely because the conditional bears
no notion of Konsequenz. So even Frege’s (very brief) discussion of
hypothetical judgment and causality bears very little relation to Kant’s
treatment of the issue.
This might just leave us with Frege’s general complaint against Kant’s
table: the reason this table can have only very little to do with Frege’s
forms of propositions is that it is governed by models of ordinary lan-
guage. Consequently, Frege’s selective approach to Kant’s table does not
merely consist in getting rid of some forms and retaining others. Rather,
it is a drastic redefinition of the forms that are retained (such as the
conditional, generality, assertion as expressed by the judgment stroke).
And this, Frege might urge, is necessary to definitively purify logic of the
psychologistic undertone it still has in Kant. But then one needs to
remember what the purpose of Kant’s table is, as opposed to the purpose
of Frege’s choice of logical constants for his propositional calculus. Frege
sets up his list so that he has the toolbox necessary and sufficient to
expound patterns of logical inference, where the truth-value of conclu-
sions is determined by the truth-value of premises, and the truth-value
of premises is determined by the truth-value of their components (truth-
functionality). Kant’s logic, on the other hand, is a logic of combination
of concepts as ‘‘general and reflected representations.’’ And we might say
that his setting up a table of elementary forms for that logic should help
us understand how the very states of affairs by virtue of which Frege’s
propositions stand for True or False, are perceived and recognized as
such. In fact, I suggest that Frege’s truth-functional propositional logic
captures relations of co-occurrence or non-co-occurrence of states of
affairs that Kant would have no reason to reject, but that for him
would take secondary place with respect to the relations of subordina-
tion of concepts that, when related to synthesized intuitions, allow us to


50
Ibid., x5, p. 14; x12, p. 27.
TRANSCENDENTAL ANALYTIC
116

become aware of those states of affairs and their co-occurrence in the
first place.
What about Frege’s challenge to the subject–predicate model of judg-
ment and his replacement of it by the function-argument model?51 Here
one might think that the modern logic of relations (n-place functions) is
anticipated by Kant’s transcendental logic, which thus overcomes the
limitations of his ‘‘general pure’’ or ‘‘formal’’ logic. For transcendental
logic is concerned not with mere concept subordinations, but with the
spatiotemporal mathematical and dynamical relations by means of
which objects of knowledge are constituted and individuated. Indeed
the most prolific of Hermann Cohen’s neo-Kantian successors, Ernst
Cassirer, advocated appealing to a logic of relations to capture the
Kantian ‘‘logic of objective knowledge’’ or transcendental logic.52
Examining this suggestion would take us beyond the scope of the pre-
sent chapter. In any event, two points should be kept in mind. The first is
that according to Kant, the relational features of appearances laid out by
transcendental logic are made possible by synthesizing intuitions under
the guidance of logical functions of judgment as he understands them.
In other words, the source of the relations in question is itself
no other than the very elementary discursive functions (functions of
concept-subordination) laid out in his table and guiding syntheses of a
priori spatiotemporal manifolds. The second point to keep in mind
is that however fruitful a formalization of Kant’s principles of transcen-
dental logic in terms of a modern quantificational logic of relations might
be, it does not by itself accomplish the task Kant wants to accomplish with
his transcendental logic and his account of the nature of categories,
which is to explain how our knowledge of objects is possible in general,
and thus explain why any attempt at a priori metaphysics on purely
conceptual grounds is doomed to fail.


51
Ibid., x9.
52
See Cassirer, Substanzbegriff und Funktionsbegriff. Peter Schulthess has defended the view
that Cassirer’s emphasis on the relational nature of Kant’s transcendental logic as well as
his emphasis on the ontological primacy of relations, not substances, is in full agreement
with Kant’s own view, including his view of logic. See Schulthess, Relation und Funktion.
Michael Friedman has defended the relevance of Cassirer’s version of neo-Kantianism for
contemporary philosophy of science: see Michael Friedman, A Parting of the Ways: Carnap,
Cassirer and Heidegger (Chicago and La Salle, Ill.: Open Court, 2000), especially ch. 6,
pp. 87–110; and ‘‘Transcendental philosophy and a priori knowledge: a neo-Kantian
perspective,’’ in Paul Boghossian and Christopher Peacocke (eds.), New Essays on the A
Priori (Oxford: Clarendon Press, 2000), pp. 367–84.
5


KANT’S DECONSTRUCTION OF THE
PRINCIPLE OF SUFFICIENT REASON




On three occasions in the Critique of Pure Reason, Kant takes credit for
having finally provided the proof of the ‘‘principle of sufficient reason’’
that his predecessors in post-Leibnizian German philosophy had sought
in vain. They could not provide such a proof, he says, because they
lacked the transcendental method of the Critique of Pure Reason.
According to this method, one proves the truth of a synthetic a priori
principle (for instance, the causal principle) by proving two things: (1)
that the conditions of possibility of our experience of an object are also
the conditions of possibility of this object itself (this is the argument Kant
makes in the Transcendental Deduction of the Categories); (2) that
presupposing the truth of the synthetic principle under consideration
(for instance, the causal principle, but also all the other ‘‘principles of
pure understanding’’) is a condition of possibility of our experience of
any object, and therefore (by virtue of [1]), of this object itself. What Kant
describes as his ‘‘proof of the principle of sufficient reason’’ is none other
than his proof, according to this method, of the causal principle in the
Second Analogy of Experience, in the Critique of Pure Reason (cf. A200–1/
B246–7, A217/B265, A783/B811).
Now this claim is somewhat surprising. In Leibniz, and in Christian
Wolff – the main representative of the post-Leibnizian school of German
philosophy discussed by Kant – the causal principle is only one of the
specifications of the principle of sufficient reason. And Kant himself, in
117
TRANSCENDENTAL ANALYTIC
118

the pre-critical text that discusses this principle, distinguishes at least
four types of reason, and therefore four specifications of the correspond-
ing principle – ratio essendi (reason for being, that is, reason for the
essential determinations of a thing), ratio fiendi (reason for the coming
to be of a thing’s determinations), ratio existendi (reason for the existence
of a thing), and ratio cognoscendi (reason for our knowing that a thing is
thus and so).1 Only the second and the third kinds of reasons (reason for
coming to be, reason for existence) are plausible ancestors of the concept
of cause discussed in the Second Analogy of Experience. Why then does
Kant describe as his proof of the principle of sufficient reason a proof
that, strictly speaking, is only a proof of the causal principle, and what
happens to the other aspects of the notion of reason or ground that Kant
discussed in the pre-critical text?
I shall suggest in what follows that in fact Kant’s response to Hume on
the causal principle in the Second Analogy of Experience results in his
redefining all aspects of the notion of reason (and, therefore, of the
principle of sufficient reason): not only the reason for coming to be
and the reason for existing (ratio fiendi and ratio existendi), but also the
reason for the essential determinations of a thing and the reason for our
knowing that a thing is thus and so (ratio essendi and ratio cognoscendi) – at
least when these notions are applied to the only objects for which one can
affirm the universal validity of some version of the principle of sufficient
reason, the objects of our perceptual experience.
In talking of ‘‘Kant’s deconstruction of the principle of sufficient
reason,’’ what I intend to consider, then, are two things. First, Kant’s
detailed analysis of the notion of ratio (reason, ground) and of the
principle of sufficient reason in his pre-critical text. Second, Kant’s
new definition, in the critical period, of all types of ratio and all aspects
of the principle of sufficient reason.2

1
These four kinds of reason, ratio, appear in Kant’s Principiorum Primorum Cognitionis
Metaphysicae Nova Dilucidatio, AAi, pp. 391–8; trans. David Walford and Ralf Meerbote, A
New Elucidation of the Principles of Metaphysical Cognition (henceforth New Elucidation), in
Theoretical Philosophy, 1755–1770. Walford and Meerbote translate the Latin ratio by
ground, and thus principium rationis sufficientis by principle of sufficient ground, which
seems odd. I have preferred to keep the term reason, and thus principle of sufficient
reason, despite the more epistemic and less ontological connotation of the term reason.
On this point, see also n. 2.
2
A point of vocabulary is in order here. The Latin term Kant uses in the 1755 New Elucidation
is ratio. In German, it becomes Grund. ‘‘Principle of sufficient reason’’ is in Latin principium
rationis sufficientis, in German Satz vom zureichenden Grund. Because the word ‘‘reason’’
appears in the ‘‘principle of sufficient reason,’’ I will use the English ‘‘reason’’ for ratio, but
PRINCIPLE OF SUFFICIENT REASON 119

One interesting result of comparing Kant’s pre-critical and critical
views is that a striking reversal in Kant’s method of proof becomes
apparent. In the pre-critical text, Kant starts from a logical/ontological
principle of sufficient reason, moves from there to a principle of suffi-
cient reason of existence (which he equates with the causal principle),
and from there to what he calls a principle of succession (a principle of
sufficient reason for the changes of states in a substance). By contrast, in
the critical text (the Second Analogy of Experience), Kant proves a
principle that looks very much like the principle of succession in the
New Elucidation, which he equates with the causal principle. And in doing
this he declares he is providing ‘‘the only proof’’ of the principle of
sufficient reason of existence and – I shall argue – he also redefines the
respective status of the ontological and logical principles themselves. In
short, instead of moving from logic to time-determination, in the critical
period one moves from time-determination to logic. This reversal of
method is related to the discovery of a completely new reason or ground:
the ‘‘transcendental unity of self-consciousness’’ as the reason of reasons,
or the ground for there being any principle of sufficient reason at all.
The discovery of this new ground has striking consequences for Kant’s
critical concept of freedom, which I shall consider at the end of the
chapter.


The principle of determining reason in Kant’s new explanation
of the first principles of metaphysical knowledge
Kant first defines what he means by ‘‘reason’’ or ‘‘ground’’ (ratio). His
definition places this notion in the context of an analysis of propositions,
or rather, of what makes propositions true.3 It is in this context that he


sometimes add ‘‘ground’’ in parenthesis, to avoid any confusion with the faculty of reason (in
German, Vernunft). In the texts from the early 1760s, logischer Grund and Realgrund are
usually translated ‘‘logical ground’’ and ‘‘real ground,’’ so in discussing these texts I shall
often switch to ‘‘ground.’’
3
By ‘‘proposition,’’ Kant means what he calls in the critical period ‘‘assertoric judgment,’’
namely a judgment asserted as true. A judgment for Kant is the content or the intentional
correlate of an act of judging. If I judge that the world contains many evils, ‘‘the world
contains many evils’’ is the content of my act of judging. It is also a proposition, a judgment
asserted as true. If I merely entertain the thought that the world may contain many evils,
without taking the statement ‘‘the world contains many evils’’ to be true, then the content of
my thought is a mere judgment, not a proposition, in Kant’s vocabulary. To move from a
mere judgment to a proposition (a judgment held to be true), one needs a reason. This,
then, is the context in which Kant defines his notion of ‘‘reason’’ or ‘‘ground.’’
TRANSCENDENTAL ANALYTIC
120

explains why he prefers to speak of ‘‘determining’’ rather than ‘‘suffi-
cient’’ reason:
To determine is to posit a predicate while excluding its opposite. What
determines a subject with respect to a predicate is called the reason. One
distinguishes an antecedently and a consequently determining reason. The
antecedently determining reason is that whose notion precedes what is
determined, i.e. that without which what is determined is not intelligi-
ble.* The consequently determining reason is that which would not be
posited unless the notion of what is determined were already posited
from elsewhere. The former can also be called reason why or reason for
the being or becoming (rationem cur scilicet essendi vel fiendi); the latter can
be called reason that or reason of knowing (rationem quod scilicet
cognoscendi).
* To this one may add the identical reason where the notion of the
subject determines the predicate through its perfect identity with it, for
instance a triangle has three sides; where the notion of the determined
neither follows nor precedes that of the determining.4

Kant gives two examples. Here is the first: we have a consequently
determining reason for affirming that the world contains many ills,
namely our own experience of those ills. But if we also look for an
antecedently determining reason, we must search for that which, in
the essence of the world, or in its relation to some other being, provides
the ground or reason for the predicate’s (for example, ‘‘containing many
ills’’) being attributed to the subject (‘‘world’’) and its opposite (say:
‘‘perfectly good’’) being excluded.
Kant’s second example is the following: we have a consequently
determining reason for asserting that light travels not instantaneously
but with an ascribable speed. This reason consists in the eclipses of the
satellites of Jupiter – or more precisely, in the delay in our observation of
those eclipses – a delay that is a consequence of the non-instantaneous
travel of light. But we also have an antecedently determining reason.
This consists, according to Kant, in the elasticity of the aether particles
through which light travels, which delays its movement.5

4
New Elucidation, x2, AAii, p. 391.
5
Kant’s view of light as a movement of fine aether particles is borrowed from Descartes. But
Kant opposes Descartes in maintaining that these particles are elastic rather than absolutely
hard, thus delaying the transmission of light (see Kant, New Elucidation, AAii, pp. 391–2,
´
and Rene Descartes, Principes de la philosophie, ed. Charles Adam, Paul Tannery, and Centre
National de la Recherche Scientifique, 12 vols. (Paris: Librairie philosophique Vrin, 1971),
part iii, xx63–4 and part iv, x28, vol. ix-2, pp. 135–6 and 215; trans. John Cottingham,
PRINCIPLE OF SUFFICIENT REASON 121

The distinction between antecedently and consequently determining
reason, as presented here, is disconcerting: clearly, the two kinds of
reason are quite heterogeneous. One is a reason for holding the propo-
sition to be true. The other is a reason for the proposition’s being true,
that is, for the state of affairs’ obtaining. Kant does recognize this differ-
ence, since at the end of his definition he characterizes the former as a
reason for knowing (ratio cognoscendi), the latter as a reason for being or
coming to be (ratio essendi vel fiendi). But he does not stress this aspect of
the distinction in his initial characterization of reasons. Both reasons are
described as reasons for the determination of a subject with respect to a
predicate. This seeming hesitation in Kant’s definition of reason
(ground) will be important for what follows.
Having thus defined the notion of reason (ratio) and distinguished two
main kinds of determining reason, Kant criticizes Wolff’s definition.
Wolff, he says, ‘‘defines reason (or ground) as that from which it is
possible to understand why something is rather than is not [definit enim
rationem per id, unde intelligi potest, cur aliquid potius sit, quam non sit].’’6 Kant
objects that this definition is circular. It amounts to saying: ‘‘Reason is
that from which it is possible to understand for what reason something is
rather than is not.’’ This circularity is avoided if one says, rather: reason
is that by which the subject of a proposition is determined, that is, that by
virtue of which a predicate is posited and its opposite is negated. That is
why it is preferable to speak of determining rather than sufficient
reason.7
But is it so clear that the Wolffian definition is circular? It is so only if
the same thing is meant by ‘‘reason’’ (in: ‘‘reason is that from which it is
possible to understand,’’ ratio est, unde intelligi potest) and by ‘‘why’’ (‘‘why
something is rather than is not,’’ cur aliquid sit potius quam non sit). But that
is not necessarily so. Wolff might have meant that the reason in the
proposition is that from which it is possible to understand the why (the
reason) in things. The parallelism of logical and ontological relations
Robert Stoothof, and Dugald Murdoch, Principles of Philosophy, in The Philosophical Writings
of Descartes, 3 vols. (Cambridge: Cambridge University Press, 1984, 1985, 1991), i , pp. 260,
270. I am grateful to Michelle and Jean-Marie Beyssade for having clarified the Cartesian
example for me.
6
Cf. Christian Wolff, Philosophia Prima sive Ontologia (Frankfurt-am-Main and Leipzig, 1736;
repr. in Gesammelte Werke (Hildesheim and New York: Georg Olms, 1962–), x56. Kant
slightly alters Wolff’s definition. Wolff actually writes, ‘‘Per rationem sufficientem intelligi-
mus id, unde intelligitur, cur aliquid sit.’’ ‘‘By sufficient reason, we understand that from
which it is understood why something is.’’ As Kant’s main discussion is about the meaning of
cur (why), however, the variation is of no consequence and we can ignore it.
7
New Elucidation, AAi, p. 393.
TRANSCENDENTAL ANALYTIC
122

would justify Wolff’s statement and dissolve the objection of circularity.
The reason Kant nevertheless formulates this objection is probably that
he shares Wolff’s view that understanding the reason in propositions
and the reason in things is really understanding one and the same thing,
the same object of intellect. But what we want to know is what is thereby
understood. Response: what is understood is what determines a subject
in relation to a predicate, that is to say, what posits the predicate and
excludes its negation.
This is where the distinction between antecedently and consequently
determining reason comes into play. But if one accepts it, then another,
more severe objection to Wolff is in order. For as we saw, Kant expressly
says that the antecedently determining reason is a reason why (ratio cur)
but that the consequently determining reason is only a reason that (ratio
quod). Given this distinction, why does Kant not make this objection to
Wolff (the reason why is not the only kind of reason), an objection that
seems, at this point, more damning than that of circularity?
This is probably because he also shares Wolff’s (and Leibniz’s) view
that the only reason worthy of the name is the antecedently determining
reason. For it is not only just a reason for our holding a proposition to be
true, but a reason for its being true. Here’s what he says on the example
of the world and its ills:
Suppose we look for the reason of ills in the world. We have thus a
proposition: the world contains many ills. We are not looking for the
reason that or reason of knowing, for our own experience plays this role;
but we are looking for the reason why or the reason for coming to be [ratio
cur scilicet fiendi], i.e. a reason such that when it is posited, we understand
that the world is not undetermined with respect to the predicate but on
the contrary, the predicate of ills is posited, and the opposite is excluded.
The reason (ground), therefore, determines what is at first indeterminate.
And since all truth is produced by the determination of a predicate in a subject, the
determining reason is not only a criterion of truth, but its source, without which
there would remain many possibles, but nothing true).8

The whole ambiguity of Kant’s position is manifest in this passage. For
on the one hand, Kant’s notion of reason (ground) is characterized as a
reason for asserting a predicate of a subject, without which there would
be no proposition susceptible of truth or falsity, that is to say, on our part,
us judging subjects, no act of asserting rather than suspending our

8
Ibid., p. 392. Emphasis in the last sentence is mine.
PRINCIPLE OF SUFFICIENT REASON 123

judgment. And the force of his statement that there must always be a
reason for determining a subject with respect to a predicate clearly rests
on the common intuition that we need a reason for holding a proposition
to be true. But understood in this way, the reason could very well be
what Kant calls a mere criterion of truth and not its source. Nonetheless,
Kant immediately adds: the reason is not simply a criterion. To deserve
the name ‘‘reason,’’ it has to be the source of the truth of the proposition.
The very same ambiguity is at work in Kant’s pre-critical proof of the
principle of sufficient reason (or of determining reason). The principle is
thus formulated: ‘‘Nothing is true without a determining reason.’’ Here,
‘‘nothing’’ clearly means ‘‘no proposition,’’ as is shown in the proof that
immediately follows Kant’s statement of the principle:
1 All true propositions state that a subject is determined with respect to a
predicate, that is to say, that this predicate is affirmed and its opposite
is excluded.
2 But a predicate is excluded only if there is another notion that, by the
principle of contradiction, precludes its being affirmed.
3 In every truth there is therefore something that, by excluding the
opposite predicate, determines the truth of the proposition (from [1]
and [2]).
4 That is precisely what is called the determining reason (definition).
5 So, nothing is true without a determining reason (from [3] and [4]).9
This ‘‘proof’’ does little more than restate what was already said in
Kant’s initial characterization of a reason: a true proposition is one in
which a subject is determined with respect to a predicate (premise [1]).
What does the determination is the reason (premise [2], and proposi-
tions [3] and [5] derived from [1] and [2]).
Consider again the proposition: ‘‘Light travels with an assignable,
finite speed.’’ To think that the proposition is true is to assert that the
predicate, ‘‘travels with an assignable, finite speed,’’ belongs to the sub-
ject, ‘‘light,’’ and that its negation, ‘‘travels instantaneously,’’ is excluded
(this is what premise [1] says). However, for such an exclusion to obtain,
there needs to be a reason (otherwise we might admit as problematic or
as possible both judgments, light travels instantaneously, light travels
with an assignable, finite speed). Now, the consequently determining
reason provided by the delay in our observation of the eclipses of

9
Ibid., prop. 5, AAi, p. 393. I have followed the progress of Kant’s argument, only ignoring a
few repetitions.
TRANSCENDENTAL ANALYTIC
124

Jupiter’s satellites excludes that the travel should be instantaneous, by
virtue of the syllogism in modus tollens: ‘‘If all light-travel is instantaneous,
there is no delay in the eclipses of Jupiter’s satellites; however, there is a
delay. So, it is not the case that all light-travel is instantaneous.’’ For its
part the antecedently determining reason excludes instantaneous travel
by the syllogism in modus ponens: ‘‘If aether particles are elastic, then all
light travel is delayed (non-instantaneous); however, aether particles are
elastic. So, all light travel is delayed.’’ The exclusion of the opposite
predicate may be derived either from the modus tollens appropriate to
the consequently determining reason or from the modus ponens appro-
priate to the antecedently determining reason.10
We see again in this example that, even if it is granted that a reason is
needed for moving from a merely problematic judgment (one with
respect to which assent is suspended) to a proposition (a judgment
asserted as true), it does not follow at all that for every truth there is an
antecedently determining reason, ratio cur. Nonetheless, just as in his
definition of reason or ground (ratio, Grund) Kant moved without any
argument from distinguishing between two types of reason (antece-
dently and consequently determining reason) to maintaining that only
one kind of reason is relevant (the antecedently determining reason,
reason for being or becoming, reason why), similarly here, Kant sub-
stitutes for the cautious conclusion that it is in the nature of propositions
(assertoric judgments) that there should be a reason for the determin-
ation of the subject in relation to the predicate (whether this reason be

10
A few quick remarks regarding my presentation of the two kinds of reasons in terms of
modus ponens and modus tollens: Kant does not explicitly give such an explanation. But the
expressions ratio consequenter determinans and ratio antecedenter determinans seem to me to be
an unambiguous reference to the idea of determining by the antecedent and determining
by the consequent of a hypothetical judgment. The corresponding logical forms are modus
ponens and modus tollens. Making this reference explicit has three main advantages. (1) We
see more clearly that the two species of ratio do not have the same force. The ratio ponens
allows us to assert universally that all light-travel is delayed (it allows us to exclude in all
cases that light-travel is instantaneous). The ratio tollens only allows us to deny a universal
judgment, excluding in this case that light-travel is instantaneous and thus allows us to
deny the universal judgment: all light-travel is instantaneous. (2) We shall see in a moment
that when Kant, just a few years later, puts into doubt the universal validity of the
antecedently determining reason, he expresses this doubt in terms of ratio ponens: he
asks, what is the synthetic ratio ponens? This confirms that his notion of reason or ground
had always been thought in light of modus ponens (or tollens). (3) In the critical period, when
Kant distinguishes a logical principle and a transcendental principle of sufficient reason,
he will define the logical principle in terms that clearly refer to the two forms, modus ponens
and modus tollens. There is thus a deep continuity in his thought on this point, which it is
important to keep in mind (see below, pp. 137–8).
PRINCIPLE OF SUFFICIENT REASON 125

antecedently or consequently determining), a far more ambitious state-
ment: there is always an antecedently determining reason for any truth:
That the knowledge of truth always demands that we perceive a reason,
this is affirmed by the common sense of all mortals. But most often we
are content with a consequently determining reason, when what is at
issue is only our certainty; but it is easy to see, from the theorem and the
definition, that there is always an antecedently determining reason or, if
you prefer, a genetic or an identical reason; for the consequently deter-
mining reason does not make truth, but only presents it.11

From this ambitious version of the principle of sufficient reason Kant
derives important metaphysical consequences that in the years to come
will motivate his growing discomfort with his own pre-critical position,
and more generally with rational metaphysics.
The first consequence of this is a proof of the principle of sufficient
reason for the existence of contingent things. This is where the concept
of cause occurs for the first time in the New Elucidation: the reason of
existence is a cause.
As a preliminary to proving a principle of sufficient reason of exis-
tence, Kant first establishes the negative proposition, ‘‘It is absurd that
something should have in itself the reason of its existence.’’12 Kant’s
proof for this proposition rests on the – unquestioned – assumption
that a cause necessarily precedes its effect in time. So, if a thing were
the cause of itself, it would have to precede its own existence in time,
which is absurd. Therefore nothing is the reason of its own existence:
Kant expressly opposes Spinoza’s notion of a God that is causa sui, cause
of itself.
On the other hand it is true to say that God’s existence is necessary, or
that the proposition, ‘‘God exists,’’ is necessarily true. But this is not
because God is the cause of himself. It is not even because his existence
is contained in his essence (as in the ‘‘Cartesian proof’’). Rather, it is
because he is the unique being that is the ground of everything possible.
I will not attempt to lay out and analyze Kant’s proof of this point. Let me
just note that, according to Kant’s pre-critical view, if we affirm the
existence of God, or if we assert the proposition, ‘‘God exists,’’ as
necessarily true, it is not by virtue of an antecedently determining
reason (whether of being, of coming to be, or of existing). There is no

11
New Elucidation, prop. 5, AAi, p. 394.
12
Ibid.
TRANSCENDENTAL ANALYTIC
126

antecedently determining reason for God’s existence, not even in God
himself. But we have a reason to assert that he exists and that this
existence is absolutely necessary. We know this by a reason for knowing
of a unique kind, which Kant will further elaborate in the 1763 text, The
Only Possible Argument in Support of a Demonstration of the Existence of God
and then thoroughly refute in the Transcendental Ideal of the first
Critique.13
Kant then sets about proving a principle of antecedently determining
reason for the existence of contingent things. The principle is: ‘‘Nothing
contingent can be without an antecedently determining reason (a cause)
of its existence.’’
The proof, roughly summarized, is the following:
1 Suppose a contingent thing exists without an antecedently determin-
ing reason.
2 As an existing thing, it is completely determined, and the opposite of
each of its determinations is excluded (definition of existence as com-
plete determination).
3 But according to the hypothesis, this exclusion has no other reason
than the thing’s existence itself. Even more, this exclusion is identical:
the very fact that the thing exists is what excludes its non-existing.
4 But this amounts to saying that its existence is absolutely necessary,
which is contrary to the hypothesis.
5 So, nothing contingent can be without an antecedently determining
reason.



13
Ibid., p. 395. The Only Possible Argument in Support of a Demonstration of the Existence of
God, AAii, pp. 83–4, trans. in Theoretical Philosophy, 1755–1770. Critique of Pure Reason,
A581–2/B609–10. The pre-critical proof rests on the idea that the notion of possible has a
‘‘formal’’ aspect (what is possible is what is thinkable, and what is thinkable is what is non-
contradictory) and a ‘‘real’’ aspect (something must be thought). Both aspects presuppose
that what is possible (thinkable) is grounded in one and the same being, which thus
necessarily exists. The Transcendental Ideal will oppose to this ‘‘proof’’ that the matter
of all possibilities, as well as the comparability of all possibilities (the formal aspect of the
possible) are provided not by an absolutely necessary being, but by the whole of reality,
given to the senses, presupposed for the collective unity of possible experience and of its
´
objects (see A581–2/B609–10, and ch. 8 in this volume, pp. 214–23). Gerard Lebrun has
convincingly shown that already in the pre-critical period, by renouncing the Cartesian
ontological proof Kant has given up the metaphysical notion of essence as a degree of
perfection and initiated instead a consideration of the conditions under which thoughts
´ ´
have meaning. See Gerard Lebrun, Kant et la fin de la metaphysique (Paris: Armand Colin,
1970), pp. 13–34. See also KCJ, p. 154.
PRINCIPLE OF SUFFICIENT REASON 127

The proof rests on three presuppositions: (a) existence is complete
determination: an existing thing is individuated by the fact that, given
the totality of possible predicates, for each and every one of them, either
it or its negation is true of the individual existing thing; (b) as such, it falls
under the principle of determining reason stated above; (c) this princi-
ple should be understood as a principle of antecedently determining
reason. If we accept all three presuppositions, then we can avoid the
absurd conclusion that a contingent existence is absolutely necessary
only if we accept that every contingent thing has an antecedently deter-
mining reason not only of its determinations (ratio essendi vel fiendi) but of
its existence itself (ratio existendi).
The second consequence of the ambitious version of the principle of
sufficient reason is a ‘‘principle of succession,’’ stated as follows: ‘‘No
change can affect substances except insofar as they are related to other
substances, and their reciprocal dependence determines their mutual
change of state.’’ Kant’s argument for this principle is that if the ground
or reason of the change of states of a substance were within it, then the
state that comes to be should always have been (given that its ratio fiendi
was always present in the substance). So, a state that was not and comes to
be must have its ground not in the substance itself but in its relation to
another substance or to other substances. (Note that this is a fundamen-
tally anti-Leibnizian view: contrary to Leibniz, according to Kant indivi-
dual substances have real influence upon one another’s states.)14
Finally, Kant devotes a fairly long discussion to the relation between
the principle of sufficient reason and human freedom. Here he opposes
a view defended by his predecessor Crusius. According to Crusius, in
some cases the existence of a state of affairs or an event is without an
antecedently determining reason. It can be affirmed only by virtue of
a ratio cognoscendi, which is none other than the existence of the state
of affairs itself as attested by experience. Such is the case with free
action: that the will should decide of its own free choice, without any

14
This principle is complemented by a ‘‘principle of coexistence’’: ‘‘Finite substances stand in
no relation to one another through their mere existence and have no community except
insofar as they are maintained in reciprocal relations through the common principle of
their existence, namely the divine intellect’’ (New Elucidation, AAi, pp. 412–13). Just as the
‘‘principle of succession’’ is the ancestor of the Second Analogy of Experience, the ‘‘prin-
ciple of coexistence’’ is the ancestor of the Third Analogy. But of course, in the Critique of
Pure Reason, as we shall see, Kant will prove both principles from the conditions of our
experience of objective time-determinations, not from the application of a previously
established principle of sufficient reason. Undertaking a detailed analysis of those two
principles and their proof in the New Elucidation is beyond the scope of the present chapter.
TRANSCENDENTAL ANALYTIC
128

antecedently determining reason, in favor of one action rather than
another, is a fact attested by experience.15 To this Kant objects that if
an action, or the will’s determination to act, were without an antece-
dently determining reason then, since the determination of the will to act
and the ensuing action have not always existed, their transition into
existence would remain undetermined – that is to say, for the action as
well as for the determination of the will, it would remain undetermined
that it should be rather than not be. Kant’s response in this case rests on
the same presuppositions as his general argument concerning the rea-
son of existence: in order to justifiably assert that a thing has come to be,
we need not only a ratio cognoscendi (ratio consequenter determinans), but
also a ratio fiendi, the ratio antecedenter determinans of its complete
determination.16
To the question: ‘‘is this principle of reason applied to human action
compatible with freedom of the will and freedom of action?’’ Kant
answers – again against Crusius – that being free is not acting without a
reason, but on the contrary acting from an internal reason that inclines
one to act without any hesitation or doubt in one way rather than
another. Kant, here, is faithfully Leibnizian.
I have suggested above that the main weakness of Kant’s argument is
the way in which he jumps from the distinction between antecedently
and consequently determining reason for asserting the truth of a pro-
position to the claim that there is always an antecedently determining
reason, a reason why. It will not be long before the universality of the
ratio cur raises doubts in Kant’s mind. But his doubts will focus at first not
on the principle of reason and its proof, but on particular cases of
connection between the ratio and the rationatum. For the analysis of
these cases, Kant introduces, at the beginning of the 1760s, the distinc-
tion between logical reason and real reason (or logical ground and real
ground), and emphasizes the synthetic character of the real ground.
When the Humean alarm clock does its work, the investigation of the
relationship of real ground to its consequences becomes generalized into
an investigation concerning the notion of reason or ground in general,
and the principle of sufficient reason itself.

15
Cf. Crusius, Dissertatio de usu et limitibus principii rationis determinantis, vulgo sufficientis
(Leipzig, 1743) (Dissertation on the use and limits of the principle of determining reason, commonly
known as the principle of sufficient reason). See also Entwurf der notwendigen Vernunftwahrheiten,
wie sie den zufaligen entgegen gesetzt werden (Metaphysik), 2nd edn (Leipzig, 1753) (Outline of the
¨
necessary truths of reason, insofar as they are opposed to contingent truths (Metaphysics)), esp. x126.
16
New Elucidation, props. 8 and 9, AAi, pp. 396–7, 398–406.
PRINCIPLE OF SUFFICIENT REASON 129


Skeptical interlude: logical reason and real reason.
The synthetic ratio ponens
In the Lectures on Metaphysics from the 1760s, Kant remarks on the
difficulty of accounting for the relation between ratio and rationatum in
the case of what he now calls ratio realis (real ground), so as to distinguish
it from ratio logica (logical ground). The logical ground (or reason), he
says, is posited by identity. But the real ground is posited without
identity. The examples show that by ‘‘real ground’’ he means the relation
of ground that connects one existence to another existence, in other words
what, in the New Elucidation, he called ratio existendi, or cause:17

All grounds (reasons) are either logical, by which the consequence is
posited by the rule of identity, where the consequence is identical with
the antecedent as a predicate. Or real, by which the consequence is not
posited according to the rule of identity and is not identical with the
ground. For instance: whence evil in the world? Response as to the
logical ground: because in the world there are series of finite things,
which are imperfect; if one seeks the real ground, then one seeks the
being that brings about evil in the world.
The connection between logical ground and consequence is clear: but
not that between real ground and consequence, that if something is
posited, something else at the same time must be posited. Example:
God wills! The World came to be. ‘‘Julius Caesar!’’ The name brings us
the thought of the ruler of Rome. What connection?18

One can find almost the same examples in the Attempt to Introduce the
Concept of Negative Magnitudes into Philosophy, which dates from the same
period.19


17
Already in the New Elucidation, Kant stressed the necessity of distinguishing between the
ground of truth and the ground of existence, that is to say on the one hand ratio essendi or
fiendi, and on the other hand ratio existendi or cause. But he did not call the former ‘‘logical
ground’’ or the latter ‘‘real ground.’’ True, he did mention the distinction made by Crusius
between ideal ground and real ground. But this distinction is not the same as the one Kant
introduced in the 1760s between logical ground and real ground. Rather, Crusius’ ideal
reason, as Kant himself points out in the Attempt to Introduce the Concept of Negative
Magnitudes into Philosophy, is what Kant calls, in the New Elucidation, ratio cognoscendi, the
ground of knowing. Cf. Kant, Attempt to Introduce the Concept of Negative Magnitudes into
Philosophy, AAii, p. 203 (trans. in Theoretical Philosophy, 1755–1770). Cf. Crusius, Entwurf,
x34ff.
18
AAxxviii, p. 12.
19
AAii, p. 202.
TRANSCENDENTAL ANALYTIC
130

In the question, ‘‘what is the connection between two distinct exis-
tences?’’ one can recognize Hume’s problem.20 But, as I have argued
elsewhere, when Kant poses the question, it is in the terms of the
Wolffian school’s logic: how are we to understand that ‘‘if one thing is
posited, another thing is posited at the same time’’? This vocabulary is
that of Wolff’s analysis of syllogisms in modus ponens. In a hypothetical
syllogism, si antecedens ponitur, ponendum quoque est consequens (if the
antecedent is posited, the consequent must also be posited).21
Interestingly, it is in the context of the modus ponens characteristic of
real ground that, it seems, Kant introduced for the first time the distinc-
tion between analytic and synthetic connection:
The relation of positing reason [respectus rationis ponentis] is connection, of
negating reason [tollentis] is opposition. The relation of logical positing or
negating reason is analytic – rational. The relation of positing or negating
reason is synthetic – empirical.22

Only with the Critique of Pure Reason does Kant think he has answered
to his satisfaction the question: what is the nature of the synthetic con-
nection between ratio and rationatum, what is the nature of the real
ground? His answer is the following: the relation of real ground, that is
to say, the necessary connection between two distinct existences, is the
connection that must necessarily exist for any order of time to be deter-
minable among the objects of our perceptual experience. But then, the
‘‘principle of succession,’’ which in the New Elucidation was a consequence
of the principle of sufficient reason, becomes the ground of its proof.
This means that the whole proof-structure of the New Elucidation is
reversed: Kant does not proceed from a principle of reason that is
both logical and ontological (every truth must have its reason, every
attribution of a property to a thing must have its reason), to a principle of
reason of existence (every contingent existence must have its reason)
and finally to a principle of succession (every change of state of a sub-
stance must have its reason in the state, or change of state, of another
substance). Instead, he now proceeds from a principle of succession (the


20
‘‘Hume’s problem’’ is Kant’s description for Hume’s skeptical doubt about our idea of
necessary connection: see Prolegomena, AAiv, p. 261. On Kant’s relation to ‘‘Hume’s prob-
lem,’’ see KCJ, p. 357, especially n. 66. And in this volume, ch. 6, pp. 147–57.
21
Christian Wolff, Philosophia rationalis sive Logica (Frankfurt and Leipzig, 1740), repr. in
Gesammelte Werke (Hildesheim and New York: Georg Olms, 1962–, ii -1), x407–8. Cf. KCJ,
p. 352. See also in this volume, ch. 6, pp. 150–1.
22
Reflexion 3753 (1764–6), AAxxvii, p. 283.
PRINCIPLE OF SUFFICIENT REASON 131

Second Analogy of Experience: ‘‘everything that happens presupposes
something upon which it follows according to a rule’’) to a redefinition of
the notion of reason or ground and, with it, to the revision of the
principle of reason in all its aspects – whether it concerns the reason of
existence, the reason of being or of coming to be, or even the reason of
knowing. It is this reversal that I would like now to examine.


The critical period: objective unity of self-consciousness
and the principle of sufficient reason
The Analogies of Experience are the principles obtained by applying to
appearances the three categories of relation: substance/accident, cause/
effect, and interaction. The Analogy now under consideration is the
Second Analogy: the causal principle, whose proof Kant takes to be
‘‘the only proof of the principle of sufficient reason.’’
Before considering Kant’s argument in the Second Analogy of
Experience, I should briefly recall three points that Kant takes himself
to have established in earlier parts of the Critique of Pure Reason, before
reaching the Analogies. The three points are the following. (1) Things as
they appear to us are perceived as having temporal determinations
(relations of succession and simultaneity) only if they are related to one
another in one time (Transcendental Aesthetic, A30/B46). (2) Things
as they appear to us are related to one another in one time only if
they appear to a perceiving consciousness aware of the unity and
numerical identity of its own acts of combining the contents of its
perceptions (Transcendental Deduction, A107/B139–40). (3) These
acts are acts of forming judgments (Transcendental Deduction,
B140–1).
By virtue of the second and third points, the reversal I described a
moment ago in Kant’s order of proof (proceeding in the critical period
from reason (or ground) of succession to reason of existence and reason
in general), is inseparable from Kant’s discovery of a new reason or
ground, one that has no precedent in his pre-critical texts (or, for that
matter, in the history of philosophy): what Kant calls the objective unity
of self-consciousness (namely the unity and numerical identity of the
self-conscious act of combining representations), which is now the trans-
cendental ground of there being any grounds, or reasons, at all, and of
the principle of sufficient reason itself.
In what follows, I will first analyze Kant’s principle of succession in the
Critique of Pure Reason, namely the Second Analogy of Experience. I will
TRANSCENDENTAL ANALYTIC
132

then show how this principle and its proof lead to a redefinition of the
reason or ground in all its aspects – reason of existence, of coming to be,
of being, and even of knowing. Finally I will show what happens to the
relation between the principle of sufficient reason and Kant’s concept of
freedom.


The proof of the Second Analogy of Experience
I have analyzed this proof elsewhere.23 I will not attempt to repeat that
analysis here, nor will I evaluate Kant’s argument in the Second
Analogy. I will consider only those aspects of it that are necessary for
our understanding of the critical notion of reason or ground, ratio.
The question Kant asks himself is well known: how do we relate the
subjective succession of our perceptions to an objective temporal order,
given that we have no perception of ‘‘time itself’’ that could provide us
with the temporal coordinates in reference to which we might determine
the positions of things or their changes of state? More specifically – this
is the problem Kant deals with in the Second Analogy – how do we relate
the subjective succession of our perceptions to an objective succession
of the states of things?
Kant’s response is in two main stages. One, fairly swift, could be
described as phenomenological. It consists in a description of our
experience of an objective temporal order. The other, longer and
more complex, rests on an argument developed earlier (in the
Metaphysical Deduction and the Transcendental Deduction of the
Categories), which concerns the role of the logical forms of our judg-
ments in establishing an intentional relation between our representa-
tions and the objects they are the representations of (or we might say, the
role played by logical forms of judgments in the directing of our repre-
sentations toward objects). I will call this second stage the logical stage of
the argument of the Second Analogy.
First, the phenomenological stage. We relate the subjective succession
of our perceptions to an objective succession of the states of things,
Kant maintains, if, and only if, we hold the subjective succession to be
determined in its temporal order. In other words, if the subjective
succession of perceptions is the perception of an objective succession,
perception A that precedes perception B cannot follow it – or rather, a
perception A0 , generically identical to perception A that preceded B,
23
See KCJ, pp. 345–75, and ch. 6 in this volume, pp. 157–77.
PRINCIPLE OF SUFFICIENT REASON 133

cannot follow perception B. To take up the well-known example Kant
uses in the Critique, perceiving that a ship moves downstream: when I
have such a perceptual experience I am aware that I could not decide
arbitrarily to reverse the order of my perceptions and, for instance,
perceive the ship again at point 1 after perceiving it at point 2. On the
other hand, if the subjective succession is only subjective, that is to say if
there corresponds to it in the object a relationship of temporal simulta-
neity, then I could, if I decided to do so, reverse the order of my
perceptions and have perception A again, or a perception A0 generically
identical to A, after having perception B (for instance – to take up again
Kant’s example – perceive the front of the house again after perceiving
the back).
One quick comment on this ‘‘phenomenological’’ stage of the argu-
ment and the examples that illustrate it. I think that the best way to
understand the description Kant proposes is to consider it as a descrip-
tion of the use that we make of our imagination in perception. When we
perceive a subjective succession as the perception of an objective succes-
sion, for instance in the perception of the ship moving downstream, at
the very moment that we perceive the second position of the ship, if we
imagine that our gaze returns to the point where we previously per-
ceived the ship, what we imagine is that we would not perceive the ship
in that place. This is what is meant by saying that the order of percep-
tions is determined. Of course, if the objective state of affairs were to
change (if we had grounds for thinking that the ship had now been
towed upstream), we could imagine that if we returned our gaze toward
the preceding point, we would see the ship again. Therefore the aware-
ness of the determined character of the order of our perception depends
not only on our senses, but also on our imagination. It is precisely
because it depends on the imagination that it can be guided both by
and toward judgment.
And this leads us to the second stage of Kant’s argument. In the first,
Kant replied to the question, how is the subjective succession of our
perceptions also the perception of an objective succession? His answer
was that this is so just in case the subjective succession is represented as
determined in its temporal order (namely, when we do not imagine that
we would perceive the same thing if our gaze were to return to the point
upon which it was focused a moment before). But this calls for answering
a second question: how and why do we hold the subjective succession to
be determined in its temporal order (why do we not imagine that
we could again perceive the same state of things at the point upon
TRANSCENDENTAL ANALYTIC
134

which we focused our gaze a moment earlier)? Here Kant’s answer
becomes more complex. I suggest that it is summed up by the following
three points. We hold the subjective succession to be determined in
its temporal order if, and only if: (1) we establish an intentional
relation between the representation and the independent object of
which we take it to be the representation; (2) in doing so, we are led to
hold the order of perceptions to be determined in the object, which
means that (3) we presuppose another objective state of things that
precedes the perceived succession and that determines its occurrence,
according to a rule. Now if this is so, we can conclude that all perceptions
of objective successions rest on the presupposition that ‘‘something pre-
cedes, upon which the perceived succession follows, according to a
rule.’’24 This ‘‘something which precedes, upon which the objective
succession follows, according to a rule,’’ is precisely what is called ‘‘a
cause.’’ It is therefore a condition of the experience of objective succes-
sions that every event (every objective succession of states in a thing)
presupposes something upon which it follows according to a rule.
But according to the Transcendental Deduction of the Categories, the
conditions of the possibility of experience are also the conditions of the
possibility of the object of experience. Therefore, it is a condition of
the possibility, not only of our experience of an objective succession, but
of that succession itself, that something should precede it, upon which
it follows according to a rule.
It would be a mistake to believe – as Schopenhauer apparently did25 –
that Kant maintains the absurd position that every objective succession is
itself a causal relation. What Kant maintains is that we perceive – that is to
say, we identify or recognize under a concept (or, more exactly, under
concepts combined in judgments) – an objective succession only if we
suppose a state of things preceding it, upon which it follows according
to a rule. For all that, we do not know this antecedent state of things. We
only presuppose it, and strive to identify it. So, for instance, perceiving
that the ship, which was at point 1, has moved to point 2, is implicitly
holding the proposition, ‘‘the ship, which was at p1, has moved to p2,’’ to
be the conclusion of a hypothetical syllogism whose major premise, and
therefore also whose minor premise, we do not know: ‘‘If q, then the


24
Cf. A189; A193/B238.
¨
25
Cf. Schopenhauer, Uber die vierfache Wurzel des Satzes vom zureichenden Grund (1813); trans.
E. F. G. Payne, On the Fourfold Root of the Principle of Sufficient Reason (La Salle, Ill.: Open
Court, 2001), ch. 4, x24.
PRINCIPLE OF SUFFICIENT REASON 135

ship, which was at p1, moves to p2; but q; therefore, the ship, which was at
p1, has moved to p2.’’ If we could not suppose the existence of something
that we could think under the antecedent q of a rule, ‘‘if q, then the ship,

<<

ńňđ. 4
(âńĺăî 10)

ŃÎÄĹĐĆŔÍČĹ

>>