. 6
( 9)


self-conscious individuals, once we were able to make our concept of
ourselves clear and distinct, it would be enough to analyze the concept
in order to know everything there was to know about me or you. It is far
from obvious that ˜˜rationalist™™ thinkers, such as Leibniz, thought of the
kind of reasoning required to see such consequences as the derivation of
analytic entailments in Kant™s sense of the word.±° However, it is also
unclear how Leibniz actually proposes to make sense of the concept of
an individual, as he understands it. For all true propositions regarding
the whole history of that individual are supposed to be entailed by the
concept of that individual. One problem is that we do not actually have
knowledge of such a complete concept for any individual. The problem
also goes deeper. For it is not clear that we even understand what such a
±° Kant and the demands of self-consciousness
concept could be. This problem is re¬‚ected in the very idea of the
identity of an individual as Leibniz understands that identity. For
Leibniz, the content that distinguishes individual simple substances
from each other consists ultimately in the di¬erential way in which they
represent other creatures representing them. With some justice, Kant
can argue that Leibniz has no real way of giving any clear non-
metaphorical sense to the di¬erent (non spatio-temporal) points of view
of those representers that Leibniz must invoke to make sense of the
di¬erences in representational content that are supposed to individuate
di¬erent simple substances. If we cannot make sense of those di¬erences
between simple substances and the di¬erent states of simple substances
in their own right, we will also not be able to make sense of the idea of a
spatio-temporal history corresponding to the di¬erent representations.
Kant thinks that Leibniz falls into this di¬culty because he attempts to
draw metaphysical conclusions from the way objects must be represen-
ted from the ¬rst-person point of view belonging to self-consciousness.
Kant is not only interested in blocking any assumptions about the
immortality of the soul that may seem to follow from my consciousness
of my identity through time, he also argues for a restriction on our
ability to have even phenomenal knowledge of the persistence of the self.
Where the numerical identity of matter is concerned, direct observation
from a third-person perspective seems to be possible. So we seem to
have knowledge of the persistence of bits of matter. Even here there are
di¬culties. The medium size material objects that are our immediate
objects of observation are persistent, but they are not themselves perma-
nent even by the lights of Kant™s theory. In the case of the self, there is a
further complication. One must already put oneself in the position of
another in order to ascribe consciousness to some other person. The
actual ascription of conscious states to the other presupposes a shift to a
third-person perspective that is itself inherently infected with the ¬rst-
person point of view. This precludes one from having any immediate
access to the other person™s existence as a persistent entity.
The only thing that one can observe to be persistent is the other
person™s body. The precise relation of that body to the other person is
quite a problematic matter. No empirical evidence seems to force us to
identify the identity of the person with the identity of that persisting
body, however plausible it is to do so. The persistence of a body is only
contingently related to the possession of a numerically identical point of
view which forms the basis for our ascription of personal identity to the
other person. Given the fallibility of memory, this means that the
The pseudo-discipline of transcendental psychology
persistence of a body may be the best criterion of personal identity that
we have, but one the satisfaction of which provides no logically compell-
ing guarantee of personal identity.
Once it becomes apparent that Kant does not intend to recognize any
public knowledge of the persistence of the self, then his failure to
introduce public time into his discussion of the role of the outside
observer begins to make sense. Nor is this failure to recognize a public
notion of inner time-consciousness an accident. Kant™s general tact is to
deny that inner experience can on its own provide any objective knowl-
edge. In order to know objective facts, we must appeal to the behavior of
physical objects in space. Our knowledge of public facts depends on our
knowledge of external objects.
° 

How independent is the self from its body?

In the last chapter I developed Kant™s general argument against the kind
of inferences that lie at the basis of rational psychology. After looking at
his general objections to e¬orts to infer substantive facts about our
nature as thinkers from the way we must think of ourselves, on the basis
of the way in which we must think of ourselves in order to ascribe
thoughts to ourselves I then turned to his arguments against treating us
as thinking things, in the sense of bona ¬de thinking substances or
persons that have identity over time. In this chapter, I explore Kant™s
arguments in the Paralogisms against Cartesian metaphysical and epi-
stemological dualism. I concentrate on the Second and Fourth Para-

  ®¤ °  ¬ §© 
The Second Paralogism attacks what Kant regards as the most illustri-
ous champion to be found among the arguments of rational psychology.
The ˜˜Achilles™™ of all rationalist proofs is the argument from the nature
of thought to the metaphysical simplicity of the thinker who thinks.
Kant™s sympathy with the argument from simplicity has to do with his
deep commitment to the logical simplicity of the ˜˜I.™™ By this he means
that our representation of self does not contain anything per se that would
serve to distinguish one self from another self. For instance in the
B-Deduction, Kant motivates the need for synthesis in order for one to
be able to represent the identity of the self in di¬erent representations by
noting that ˜˜through the I as simple representation no manifold is
given™™ ( ±µ).± In that context, Kant contrasts the kind of simplicity of I
representation that is supposed to characterize our discursive minds
with the kind of intuitive intellect that could be ascribed to God.
Neither Descartes nor Leibniz would disagree with Kant™s theory that
rational psychology would have to be based on ˜˜the simple and for itself
How independent is the self from its body?
completely empty of content representation: I™™ ( / °). The
simple representation ˜˜I™™ is the subject of consciousness represented as
subject of self-consciousness. It is simple because it lacks any content
beyond its self-referential form. Thought seems only to be possible if the
thinker is characterized as a thinker by simplicity or absolute unity of
consciousness. One cannot divide distinctive thoughts among di¬erent
subjects without depriving those thoughts of their very existence. A
verse or a sentence cannot be broken up into its constituent words and
its constituent words represented by di¬erent individuals and still be that
particular thought. From the inherent unity of particular thoughts it
seems to follow that thought requires a metaphysically simple, or abso-
lutely unitary subject. But Kant points out that the inference to the
existence of an absolutely unitary bearer of thought is based on a
presupposition that deprives it of most of its force.
We must presuppose a single logical subject, or subjective ˜˜I™™ ( µ)
in order to ascribe to ourselves any thought. The logical subject lacks
any content aside from that provided by that of which one is conscious.
In this sense, the statement ˜˜I am simple™™ is an immediate expression of
self-consciousness ( µµ). But the fact that we can only represent
ourselves as thinking by attributing unity to each of our points of view as
thinkers does not establish that we are beings that have an absolutely
unitary or simple nature.
Whenever we represent ourselves as the subject of di¬erent experien-
ces, regardless of whether we are the same individual bearer that had all
of those experiences, we represent ourselves as what Kant calls ˜˜the
logical subject of a thought.™™ The logical subject of a thought is just the
subject that is represented by us as the thinker of that thought. The unity
of the point of view does not guarantee the unity of the bearer of that
point of view. Thus, an individual who had undergone ¬ssion, fusion, or
had his or her experiences transplanted from those of another individual
could still regard him- or herself as the same logical subject of all those
experiences even though there was no individual bearer that had all of
those experiences.
Thus, even many simple substances might again ¬‚ow into one, without any-
thing more being lost than the greater amount of subsistence, the one contain-
ing the degree of reality of all the previous ones together in it, and perhaps the
simple substances that provide the appearance of matter could bring forth
children souls through dynamic division of parent souls as intensive magnitudes
(admittedly not through mutual mechanical or chemical in¬‚uence, but rather
through an in¬‚uence that is unknown to us of which they are the mere
± Kant and the demands of self-consciousness
appearance), while they might in turn replenish their loss through coalition
with new stu¬s of the same kind. ( ±·n)

While the self presents itself to itself in self-consciousness as a unity that
seems to be the unity of a real particular, this unity turns out to be, at
least in some cases, a mere logical unity belonging to all the thoughts
which that subject thinks of as belonging to him- or herself. Whether
that subject is a thing with real unity, as well, is a matter for further
Self-consciousness involves a reference to self in abstraction from any
description which distinguishes one from other individuals. Self-con-
sciousness thus encourages us to think that the real subject of thought is
simply identical with the logical subject of self-conscious thought. The
logical subject of thought is just the point of view to which we ascribe a
certain unity. From the intrinsic unity of a point of view, which Kant
refers to as logical simplicity, one cannot establish the intrinsic unity or
simplicity of the real bearer of that point of view. For one would have to
establish the identity of the real with the logical subject of thought in
order to infer that the bearer of thought is simple because the represen-
tation of that bearer is simple. The logically simple subject to which we
ascribe individual thoughts might turn out itself to have an intrinsic
unity of point of view that supervenes on a complex of many diverse real
bearers. The proposition that a thought can only be the expression of
the absolute unity of a thinking being is thus not analytic ( µ). It
involves a substantive claim about the nature of thinking beings. Nor
can experience alone provide the kind of evidence required to support a
claim to necessity or absolute unity.
Kant thinks of the transcendental subject as the bearer of thought
when it is conceived in abstraction from any of the features which
distinguish one thinking being from another: ˜˜It means a something in
general (transcendental subject) the representation of which must in-
deed be simple, precisely because one certainly cannot think of anything
more simple than by means of the concept of a mere something™™ ( µµ).
But Kant insists that the simplicity of our representation of ourselves is
not the simplicity of a real bearer of thought. Transcendental self-
consciousness is the ˜˜I™™ attached to thought that ˜˜merely transcenden-
tally designates™™ the real bearer of thought ( µµ). It refers to the bearer
of thought only insofar as that bearer is thought of as the condition for
the possibility of thought, but not to any further features of that bearer.
In self-consciousness, one refers to oneself where ˜˜oneself™™ is merely
How independent is the self from its body?
serving as the purely formal second term of a re¬‚exive relation which
picks up the reference involved in the use of the other term of that
re¬‚exive relation.
The plausibility of thinking of the self as simple rests on one™s need to
take the ¬rst-person perspective in order to ascribe thoughts. The
¬rst-person perspective carries over to the ascription of thought to other
persons. In the case of other rational beings, one must put oneself in the
position of those thinking beings in order to be able to ascribe thoughts to
them as well. Kant again insists that we can only represent another
thinking being by transference from our own self-conscious re¬‚ection. He
then explicitly rejects the inference as invalid that the self must therefore
be simple. We are wrong in moving from the fact that self-consciousness is
a necessary subjective condition under which we can make the notion of a
rational being intelligible to ourselves, to the conclusion that the unity that
we ascribe to ourselves as subjects of thought is a constitutive feature of the
concept of a thinking being in general:
Thus here, as in the previous Paralogism, the formal proposition of appercep-
tion: I think, is the whole ground on which rational psychology dares its
extension of its knowledge, which proposition is indeed no experience, but the
form of apperception, which belongs to every experience, and precedes it, yet
must only always be regarded in respect to possible knowledge as merely subjective
condition of it [knowledge], which we unjustly make into the condition of the
possibility of knowledge of objects, namely into a concept of a thinking being in
general, since we cannot represent it [a thinking being] except by putting
ourselves with the formula of our consciousness in the place of every other
intelligent being. ( µ)
The form of apperception, the ability to represent experiences in
terms of an ˜˜I think™™ through which we are conscious of ourselves being
conscious of those experiences, is the basis for all experience (at least of
intelligent beings, as we can understand them). Apperception is the basis
for our representation not only of ourselves, but of other rational beings.
Yet it is a subjective condition that we cannot construe as the objective
basis for knowledge of de re necessities concerning the nature of rational
beings. The only grip we have on what it is to be a thinking being is
based on our ability to understand what it would be like for us to
perceive or understand the world in a di¬erent way than we do. But it
does not follow from this that the very nature of a rational being involves
a connection to self-consciousness. The point of view that we attribute to
thinking beings on the basis of our self-consciousness must be inherently
unitary and indivisible, but we cannot legitimately conclude from this
± Kant and the demands of self-consciousness
subjective necessity governing our ascription of thought that rational
beings must have an inherently unitary nature.
In the second edition of the Paralogisms ( °) Kant seems at ¬rst to
endorse a more metaphysically loaded notion of the self™s simplicity
which suggests a much greater sympathy with rationalist critics of
materialism. He notes that apperception is something real whose sim-
plicity lies in its very possibility. The ¬rst-person point of view of
self-consciousness demands that the subject of self-consciousness be a
single unitary subject. Kant assumes as an independent premise that
there is nothing in space that is both real and simple. He concludes from
these assumptions that my character as a mere thinking subject cannot
be explained by appeal to materialism. The assumption that what is
spatially real cannot be simple expresses his thesis that points are
abstractions from the real extended constituents of space together with
the additional assumption that points are the only simple things that
have spatial location. Both of these assumptions are problematic, since
they involve a rejection of the existence of point-events.
The argument has some force against the very crude reductive forms
of materialism with which Kant was familiar (Helvetius for instance),
although it depends on an assumption about the nature of space that is
now quite controversial. But even if we waive the worry that the
argument has no force against someone who maintains that matter
could be composed out of non-extended point-objects, it does not o¬er a
very powerful objection to non-reductive materialism, which is pre-
pared to argue that mental events are not reducible to physical events.
Nor does the argument pose a problem even for a sophisticated form of
eliminative materialism that treats the subjective point of view as some-
thing that has instrumental signi¬cance in understanding experience,
but no ultimate real existence. A non-reductive materialist or even an
eliminative materialist can allow for the existence of an emergent
property of absolute unity or simplicity that cannot be adequately
understood in terms of any of the individual bits of matter that collec-
tively have this emergent property.
Despite his rejection of materialism, Kant admits that the appeal to
the simplicity of self-consciousness cannot establish the thesis of spiritu-
alism or mentalism, that is, the independent existence of mental events,
since the manner in which I exist is not determined by self-consciousness
per se. Indeed, this is what is to be expected given the overall critical
attitude to Cartesianism that underlies the argument of the Paralogisms.
In the Cartesian tradition, the simplicity of the self as thinker has been
How independent is the self from its body?
used to support arguments for the immateriality of the soul. At ¬rst
blush, Leibniz seems to be an obvious target for Kant™s critique. Thus
Leibniz argues in Monadology, section ±·, that mental states cannot be
explained in terms of mechanical composition, but have a unity that can
only be understood by recourse to simple substances. Kant agrees with
Leibniz that mechanism fails to explain representation, but he does not
accept Leibniz™s inference to the existence of simple substances. On the
other hand, Leibniz argues for spiritualism from the simplicity of self-
consciousness, not for immaterialism. In fact, he does not think that the
soul or self can ever exist in a completely disembodied state (Monadology,
section ·).
Although Kant directs his argument primarily against the Leibniz“
Wol¬an school of thought, Descartes is a better target. For Descartes
does think that the soul is not only simple, but, in principle, can exist in a
disembodied state. The status of embodiment in Descartes is, however,
more complicated than it at ¬rst appears to be. In the Sixth Meditation,
he rejects the idea that the soul is related to the body as a pilot to a ship.
He argues that the human being qua embodied soul has the complex
unity of an individual even though body and soul are di¬erent substan-
Now, Kant argues that the statement ˜˜I exist as thinking™™ involves an
element of experience. But, in order to show that I am a substance in
any but a purely formal sense, I would have to show that it is in my
nature to be an object that persists over time. Unfortunately, a persistent
object cannot be found in inner experience alone, and hence I have no
reason to assume that the formal identity of the self as subject of the
experiences that I ascribe to myself or to others provides adequate
grounds for making substantive claims about my relation as thinker to
my body or other bodies. Kant concludes that no valid inference can be
drawn concerning either the dependence or independence of the soul
from the body. That Kant thinks that no inference can be drawn
concerning the relation of consciousness to the body is only con¬rmed
by a look at a parallel passage from ˜˜On the Progress of Metaphysics,™™
Ak. , p. °:
That he [the human being] is not wholly and purely corporeal may be strictly
proven, if this appearance is considered as a thing in itself, from the fact that the
unity of consciousness, which must be met with in all cognition (including that
of oneself ) makes it impossible that representations divided amongst various
subjects could constitute a uni¬ed thought: therefore materialism can never be
used as a principle for explaining the nature of the soul.
± Kant and the demands of self-consciousness
Kant ¬rst argues that the claim that human beings are not wholly and
purely corporeal can be proved from the fact that the unity of conscious-
ness cannot be distributed among di¬erent subjects, if the human being as
appearance is considered as a thing in itself. But the whole point of the
Paralogisms is that the antecedent of the conditional is something that
we cannot take for granted. It is precisely the mistake of the rationalist
and materialist philosophers that Kant wishes to criticize that they take
what appears to us to be what that thing is in itself. But Kant argues that
we cannot infer from the subjective conditions under which we can
alone recognize other minds that all rational beings in general must be
such as those constraints dictate to us ( µ).
In the ˜˜Progress™™ draft he then suggests that we can distinguish the
body from the soul as phenomenon and still maintain that the external
thing (the basis for the phenomenon of body) could itself be a simple
being when taken as a thing in itself. In short, if we are not entitled to
take the unity of consciousness in appearances as a unity of conscious-
ness in things as they exist in themselves, the argument does nothing to
contradict materialism. Given the conditional nature of Kant™s argu-
ment and his own skepticism about the truth of the antecedent, I think
one must reject Henry Allison™s view that Kant takes this to be a
compelling argument against materialism.µ The argument provides the
basis for rejecting the possibility of materialism only if we take the
standpoint of the transcendental realist who maintains that inferences
from the mental unity of appearances to the mental unity of things in
themselves are legitimate. Thus, the argument has at best ad hominem
force against a materialist who accepts transcendental realism.
How does Kant reconcile his rejection of both spiritualism and
materialism as accounts of the ultimate nature of human beings? In our
experience, the mental and the physical are distinct, since the mental is
that which is essentially inner, and the physical is that which is essential-
ly outer relative to the point of view of the representations which make
up the mental.This leads Kant to favor property dualism, the view that
mental and physical properties are distinct kinds of properties, at the
phenomenal level of our experience. In the Paralogisms, Kant does not
actually defend phenomenal substance dualism, the view that there are
distinct mental things and physical things. However, he does distinguish
objects of inner sense from objects of outer sense. And, in the Metaphys-
ical Foundations, he maintains that life involves action that can only be
understood in terms of desires that do not belong to outer sense. He
concludes that ˜˜when we look for the cause of any change of matter in
How independent is the self from its body?
life, then we must immediately look for it in another substance that is
however connected to matter™™ (Ak. ©, p. µ). Kant™s claim that we
must look to another substance as cause of change in life as opposed to
the substances that cause change in matter strongly suggests that he was
committed, at least at that time, to phenomenal substance dualism.
While Kant seems to be attracted to some form of substance dualism
at the level of phenomena, he does not ¬nd such a dualism defensible at
the level of things as they are in themselves. Such substance dualism
would be forced on us if we were to treat the distinction between matter
and mind as ultimate, as a feature of things as they are in themselves.
But we cannot do this due to the limitations of what we can infer from
the conditions under which we ascribe thoughts to individuals. Kant
thus allows for the possibility of a form of naturalism at the level of things
as they are in themselves which gives up substance dualism, and perhaps
even property dualism, at the most fundamental level of reality:

If matter were a thing in itself, then it would be totally di¬erent as a composite
being from the soul as a simple one. Now it is only external appearance whose
substratum cannot be known through any predicates that can be given; hence I
can well assume that it [the substratum] is in itself simple, although in the way
that it a¬ects our senses, it generates the intuition of the extended and hence of
the composite. And I can assume that thoughts which can be represented with
consciousness through its own inner sense belong to the substance in itself
which in respect to our external senses has extension. In this way, precisely the
same thing which in one respect is called corporeal would be in another a
thinking being whose thoughts we cannot intuit, but whose signs in appearance
we can intuit. Thus the expression would lapse that only souls (as particular
kinds of substances) think; it would be said, as is customary, that human beings
think, i.e. that precisely the same thing which is extended as external appear-
ance is internally (in itself ), a subject that is not a composite but rather simple
and thinks. ( µ)

Kant argues that instead of talking about thinking substances, or souls
that think, it might turn out to be more appropriate to treat of human
beings as the bearers of thoughts as ordinary language does. In e¬ect, he
seeks a common ground of the mental and the physical or rather a
concept of substance which is su¬ciently rich to be able to explain both
mental and physical properties in experience. This is a rather attractive
position. Characteristically, Kant does not think that a substance that
would have mental as well as physical properties is knowable by us (
°). We have knowledge of matter and of our inner states. But matter,
such as we know it, is a mere appearance, as are our inner states. There
±° Kant and the demands of self-consciousness
is something transcending appearance which is responsible for our
representations of inner states as well as of material states. We do not know
that there is an intrinsic di¬erence between what would explain the
properties of matter and what would explain mental events. At the
deepest level, Kant is therefore suspicious of spiritualism and its claim
that the mind is ultimately either more basic or independent of the
physical, without being any more sympathetic to physicalism. Given his
strictures on our capacity to know how things are in themselves inde-
pendently of the conditions under which we can know them, Kant is
also unwilling to endorse the hypothesis that there is a kind of common
substance that underlies both mental and physical appearances.

   ¦  µ   °    ¬ § ©  
So far we have looked at three di¬erent determinations of self-con-
sciousness that suggest rich metaphysical commitments. Self-conscious-
ness presents itself to us, and hence also a thinker to us, as a logically
basic subject that has identity over the di¬erent thoughts that it ascribes
to itself. The subject of thought, and hence thinkers as they present
themselves to us, has an intrinsic unity or simplicity to it that seems to
distinguish them from material objects. In self-consciousness, I also
distinguish my existence as a thinker from other things outside of my
consciousness, including my own body. This suggests that I might have
an existence as a thinker that is somehow independent of my body. But
Kant rightly insists that the truth of the claim that I distinguish myself
from other things in self-consciousness is analytic ( °). It does not
establish anything about the ultimate nature of my relationship to other
In the Fourth Paralogism in the ¬rst edition, Kant argues against the
idea that the existence of things that are external to the self is doubtful.
Descartes, for instance, famously argues in the Second Meditation that
the existence of things outside of the self is doubtful in a way that the
existence of thinking things is not. Descartes bases his real distinction
between minds and bodies on the certainty of our knowledge of our
mind and the uncertainty of our knowledge of (even our own) body. In
the First Paralogism, Kant already rejects Descartess inference from the
privileged epistemic status that thinkers have to the claim that they are
thinking substances. In the Second Paralogism, he then rejects the claim
that thinking things are, or even must somehow be, really distinct
substances from material or extended things. In the present context,
How independent is the self from its body?
Kant™s concern is with rejecting attempts to infer that things outside of
us have a di¬erent epistemic status than things as they are presented to
us in thought.
Kant™s general argument is directed against what he calls idealism. It
is rather surprising to ¬nd a philosopher refuting idealism as something
scandalous and vaguely pernicious, when he characterizes his own
philosophy as a ˜˜transcendental idealism.™™ Although Kant does refer to
his own philosophical position as a form of idealism, he identi¬es
idealism tout court (without predicative modi¬ers) with the denial of the
existence of mind-independent objects. This is apparent in the
Idealism consists in the assertion, that there are none but thinking beings, all
other things, which we think are perceived in intuition, being nothing but
representations in the thinking beings, to which no object external to them
really corresponds. Whereas I say, that things as objects of our senses existing
outside of us are given, but we know nothing of what they may be in themselves,
knowing only their phenomena, that is, the representations which they cause in
us by a¬ecting our senses. Consequently, I grant by all means that there are
bodies without us, that is things which though quite unknown to us as to what
they are in themselves, we yet know by the representations which their in¬‚u-
ence on our sensibility provides us with, and which we call bodies, a term
signifying merely the appearance of the thing which is unknown to us, but not
therefore less real. Can one really call this idealism? It is the very opposite of
it. (section ±, Ak. ©, pp. “)

Kant maintains that skepticism about the existence of external objects
is generated by a commitment to transcendental realism. This transcen-
dental realism is the thesis that objects as they (veridically) appear to us
are things as they exist in themselves. Transcendental idealism, by
contrast, maintains that we cannot know what things are in themselves.
We can know facts only about appearances. Somewhat puzzlingly,
Kant identi¬es transcendental realism with what is usually referred to as
representational realism. For he maintains that the transcendental real-
ist must assume that we have direct perceptions only of what is represen-
ted by us as represented by us. The direct or naive realist is inclined to
reject any need to assume the existence of such epistemic intermedia-
ries. The need to assume epistemic intermediaries is usually motivated
by the need to provide an account of sensory illusion and other delusive
experiences such as dreams. The representational realist insists that
delusive and veridical perceptions should be treated symmetrically. If
we do not assume any epistemic intermediaries, it is hard to understand
± Kant and the demands of self-consciousness
how delusive experiences could occur. And, once we assume the exist-
ence of such intermediaries in the case of delusive perceptions, it is hard
to see why such intermediaries simply drop out in the case of veridical
Although the argument of the representational realist has some
plausibility, it is hard to ¬nd it compelling. For it is open to the direct
realist to argue that things as they are in themselves are just the way
things appear to standard observers under standard circumstances,
while delusive appearances are just the way things appear to standard or
non-standard observers under non-standard circumstances. Once one
makes this move, one must then spell out what standard observers and
standard circumstances are supposed to be. The most plausible way to
do this is in terms of the operation of certain causal mechanisms. The
skeptic can then argue that, in principle, we are unable to know whether
the causal mechanisms which cause standard observers under standard
circumstances to perceive things as thus and such are in fact ever set up
in the manner which we believe them to be. In this case, we will be
systematically deluded about the world. We ¬nd ourselves forced to
admit the possibility that empirical idealism could be true, in other
words, that all of our beliefs about the external world might turn out to
be false.
The key to an argument against empirical idealism within the frame-
work of Kant™s transcendental idealism is the combination of immediate
consciousness of outer objects with a denial that the immediate relation
to objects outside of one™s subjectivity communicates (intuitive) knowl-
edge of the things with which one is in an immediate relation. Kant
identi¬es idealism with skepticism about our ability to justify claims
concerning the external world. This is somewhat puzzling at ¬rst, since
we are used to thinking of such a position as a consequence of metaphys-
ical realism. But it turns out that idealism is to be understood as
empirical idealism, and Kant thinks of such empirical idealism as a
consequence of what is now generally called metaphysical realism, but
which he refers to as transcendental realism. Empirical idealism is to be
distinguished from empirical realism. The distinction between the ideal
or mind-dependent and the real or mind-independent is a distinction
which falls here within the domain of experience and the objects
belonging to experience.
According to empirical realism, we have immediate perceptual
knowledge of objects which are external to us. Kant also refers to
empirical realism as dualism. We have direct perceptual knowledge of
How independent is the self from its body?
things, objects in space, which are not minds, and knowledge of inner
states which are inherently mental, as well. Hence the term ˜˜dualism.™™
Kant™s use of the term ˜˜dualism™™ does not at ¬rst seem to suggest a
commitment to substance dualism, but rather to a form of property
dualism according to which mental and physical states are essentially
distinct, at least in what appears to us in experience. However, he does
conclude from the di¬erent character of the objects which we directly
perceive to be temporal and spatial that inner and outer states are to be
ascribed to di¬erent substances. He thus explicitly commits himself to
substance dualism with respect to phenomena, that is, objects as we
must experience them ( ·).
Kant is convinced that empirical idealism is a consequence of a
commitment to transcendental realism ˜˜which regards time and space
as something given in themselves, independently of our sensibility™™ (
). If space and time and the objects given in space and time are
radically independent of our minds, then there seems to be no way of
establishing that those objects must be as we think them to be. The
transcendental or metaphysical realist must always allow for the possi-
bility that all of our beliefs could be false. The transcendental idealist,
by contrast, and Kant subscribe to transcendental idealism, and main-
tain that space and time are nothing apart from the sensible conditions
under which we represent objects. Thus there would be at least no
spatial objects to which our beliefs fail to conform so long as they
satisfy the best standards we can have for the determination of
whether our beliefs about objects external to us in experience are true.
This raises more questions than it answers. For, in order to get a
signi¬cant contrast between transcendental realism and transcendental
idealism, Kant takes transcendental idealism to deny not that things
existing independently of our sensibility could exist at all, but only that
they are knowable for us. This seems to push the problem posed by
idealism with respect to external objects back to the problem of how
objects that are external to us in space relate to objects that are
external to us in a sense that is independent of the way in which we
experience them.
Even if the transcendental realist™s attempt to provide an objective
account of subjectivity is doomed to failure, this does not by itself rescue
Kant™s position from the pressures which threaten it. In the more
extensive ¬rst edition discussion of the Fourth Paralogism, Kant seems
to refute the skeptic only by granting to him or her everything that s/he
± Kant and the demands of self-consciousness
I do not need to draw an inference with respect to the reality of external objects
any more than I do with respect to the reality of the object of my inner sense (of
my thoughts), for they are both nothing but representations the immediate
perception (consciousness) of which is also a su¬cient proof of their reality. (

While Kant insists that we have a direct perception of objects in a space
which is outside of us, it turns out that this space and the bodies in it are
mere representeds, ˜˜the objects of which are something only through
these representations. Apart from them they are nothing™™ ( ·°). This
refutation of idealism has nothing to recommend itself over Leibniz™s
most phenomenalist claims:

I judge without proof, from a simple perception or experience, that those things
exist of which I am conscious within me. There are, ¬rst myself, who am
thinking a variety of things, and then, the varied phenomena or appearances
which exist in my mind. Since both of these are perceived immediately by the
mind without intervention of anything else, they can be accepted without
question, and it is exactly as certain that there exists in my mind the appearance
of a golden mountain or of a centaur when I dream of these, as it is that I who
am dreaming exist, since both are included in the one fact that it is certain that a
centaur appears to me.

Neither Leibniz nor Kant leave the matter at such immediate conscious-
ness of objects.· They both see the need to distinguish between real and
imaginary objects by means of coherence considerations. Thus, Kant
claims that we can distinguish illusory appearances from bona ¬de
perceptions of external objects by appeal to the principle that ˜˜what is
connected with perception according to empirical laws, is real™™ ( ·). This is also
a claim Berkeley could endorse. Thus Kant™s refutation of idealism
seems to have nothing to commend it over Leibniz™s or Berkeley™s
similar arguments against skepticism about the existence of bodies.
Indeed, refutations of idealism were standard in the handbooks of
Leibnizian“Wol¬an philosophy which Kant used for his own lectures.
Kant™s ¬rst critic, Garve, or rather Feder, can hardly be blamed for
assimilating Kant™s position to Berkeley™s phenomenalism. Like Kant,
Berkeley thought that the best response to the skeptic about external
bodies was to argue that the only intelligible notion of bodies we have is
one essentially tied to our sense representations. In responding to the
criticism that his view was indistinguishable from Berkeley™s phenom-
enalism, Kant came to reformulate his criticism of empirical or psycho-
logical idealism in the second edition of the Critique. He retains the idea
How independent is the self from its body?
that we have a direct perception of bodies, but he gives up the idea that
these bodies are mere internal accusatives of perception.
In the next chapter, I turn to Kant™s argument against Cartesian
skepticism about the existence of external objects in the second edition
Refutation of Idealism. Discussion of the Refutation allows me to ¬‚esh
out the implications of my overall argument concerning the conditions
under which our consciousness of ourselves as temporal and spatial
beings is possible. I focus on the manner in which our temporal con-
sciousness of ourselves as distinct individuals with distinct experiences is
parasitic on our experience of objects that exist outside of us.
The Refutation of Idealism draws on the full resources of the analysis
of the role of causation, and especially substance, in determining the
temporal relations between inner episodes. For, in the Refutation, Kant
argues that the mental lacks the autonomy required to fund the notion
of persistence required to make sense of the very ascription of determi-
nate temporal episodes to oneself needed in order for one to make sense
of inner experience. From there, Kant argues that we cannot have any
determinate beliefs about our inner states at all without also having
beliefs about objects that exist outside of us. These objects turn out in the
end to be objects that are outside of us in the transcendental, as well as
the empirical, sense. Here, Kant fully exploits the idea that the inner“
outer distinction is constitutive of any determinate consciousness of self
by showing how the inner“outer distinction constitutes the particular
consciousness of self that we have as individuals who have distinct
experiences that, in principle, must be subject to some temporal order if
they are to be comprehensible to us at all.
° ±°

The argument against idealism

O¬cially, the Refutation of Idealism is the only novel addition to the
Critique in its second edition ( ©n). Given the other changes that
Kant makes in the second edition, this admission may justify the amount
of attention that has been devoted to such a small amount of text. He
piques the interest of other philosophers by intimating that he has solved
a problem that has been a ˜˜scandal™™ to philosophy and all reasoning

Idealism may be held to be ever so innocent with respect to the essential
purposes of metaphysics (which it in fact is not) it still remains a scandal of
philosophy and of the universal reason of humanity that the existence of things
outside of us (from which we however derive the whole stu¬ of cognitions even
for our inner sense) must be accepted on faith, and not to be able to o¬er a
satisfactory proof to someone if it should occur to him to doubt it. ( ©)±

Since Kant advocates his own form of idealism, transcendental ideal-
ism, the Refutation is not directed at all forms of idealism. The Refuta-
tion is directed against a speci¬cally modern, and post-Cartesian skepti-
cal worry about whether beliefs about outer states have the same degree
of warrant as beliefs about inner states. The Refutation responds to the
provisional skeptical position outlined in the ¬rst and second of De-
scartes™s Meditations on First Philosophy, but Kant™s more immediate target
is the Humean position of his contemporary, Friedrich Jacobi. In the
passage above, Kant alludes to Jacobi™s claim that we need to appeal to
faith or belief (Glaube) in order to support claims about the existence of
the external world.
Where Descartes worries about the possibility that my beliefs about
external objects might all turn out to be false, Kant wants to show that
any knowledge of inner states entails the existence of objects existing
outside of me. In this way, he also hopes to show that Jacobi™s claim that
The argument against idealism
we have only an unjusti¬able belief in the existence of external objects is
unfounded. Kant aims to show not only that we have no reason to doubt
the existence of the external world, but that we must assume its existence
in order to have any determinate inner experiences at all.
The Refutation has a particular interest for my own enterprise of
articulating the conditions under which a distinctive consciousness of
ourselves is possible. Kant argues here that we can only know what is
internal to our own individual experiences insofar as we are directly
conscious of something that is external to our own individual experi-
ence. The Refutation thus links knowledge of oneself as an empirical self
with the inner“outer distinction, while arguing that our consciousness of
what is internal to our own distinctive point of view in experience is
parasitic on what is outside of it in experience. In the Refutation itself,
Kant is concerned only with the relation in experience between what is
internal to a point of view and what is external to a point of view, but, in
later personal notes, he then seeks to ground the distinction between the
inner“outer within experience on the distinction between what is inter-
nal to experience and what is external to it. In this way, he seeks to show
that any consciousness of oneself as an experiencer involves the empiri-
cal, as well as the transcendental distinction between the inner and the
In responding to Descartes™s problematic ˜˜material™™ or ˜˜psychologi-
cal™™ idealism, Kant argues that Descartes is wrong when he takes inner
experience to be indubitable and the experience of outer objects to be
dubitable ( ·µ). According to Descartes, one has indubitable knowl-
edge of any mental state that one is in purely in virtue of being in that
state, while one™s beliefs about outer objects are dubitable. This is the
basis for the real distinction that he attempts to draw in the Second
Meditation between mental and physical states and their respective
bearers. Kant argues, by contrast, that one cannot even have knowledge
of inner episodes except on the basis of a belief in the existence of outer
objects that cannot be false. Kant does not address or even mention
Descartes™s appeal in the Fifth Meditation to the bene¬cence of God to
warrant the reliability of our assumption that there is an external world.
Kant would, however, reject such an appeal as based on knowledge of
God that we do not have, and that is at any rate less reliable than our
knowledge of the existence of an external world.
Kant™s aim in the Refutation is to prove that experience would be
impossible if there were no bodies and nothing that exists outside of
consciousness. He has no argument against a person who is skeptical
± Kant and the demands of self-consciousness
about having any experience or self-knowledge at all. This leaves a kind
of gap in Kant™s argument. He cannot refute a Cartesian skeptic who
limits him- or herself to the cogito. Kant does ¬rst note that Descartes™s
˜˜problematic idealism™™ declares ˜˜only one empirical assertion (assertio),
namely: I am for indubitable™™ ( ·), and Kant does set out to refute
˜˜problematic idealism.™™ However, he soon goes on to describe his aim
in the Refutation as that of showing that ˜˜even our inner experience, not
doubted by Descartes, is only possible under the condition of external
experience™™ ( ·µ). And Kant later notes that the representation ˜˜I am™™
is not experience, since one needs more than mere existence for experi-
ence, one needs some more determinate representation of the individual
in time ( ··).
However, even if the argument has no force against a skeptic who is
willing to retrench and restrict him- or herself to the certainty of the
cogito, the argument is still quite ambitious. Kant attempts to show that
one can only have a consciousness of oneself as existing in time, and
hence the capacity to justify one™s beliefs about one™s own inner states, if
one has some true beliefs about the existence of objects existing outside
of one. He starts from the assumption that I am conscious of my own
existence as determined in time. He then appeals to a conclusion of the
First Analogy as an independent premise: all time-determination pre-
supposes something permanent in perception (relative persistence is
actually enough for the present purpose). What he wants to establish is
that what persists cannot be an intuition in me. Representations make
up the only bases in me for determining my existence. And as represen-
tations they themselves also require something persistent that is distinct
from them. For something is needed in relation to which representations
may be said to change and in relation to which my existence in time may
be determined.
Two conclusions are supposed to follow: (±) the perception of what is
persistent or permanent is only possible through a thing existing outside
of me and not through the mere representation of a thing outside of me;
() the determination of my existence in time is only possible through the
existence of actual things that I perceive outside of me. Since the
Analytic as a whole has argued that consciousness in time is necessarily
tied to the possibility of time-determination, Kant can then also avail
himself of this claim as a premise from which he draws the ¬nal
conclusion of his argument: () since the existence of things outside of
me is a condition for time-determination, consciousness in time is
necessarily tied to the existence of things outside of me. This is just
The argument against idealism
another way of claiming that consciousness of my own existence is also
an immediate consciousness of things that exist outside of me.

µ ® ¤      ® ¤ © ® §     § µ  ® 
The Refutation of Idealism links empirically determined self-conscious-
ness to the existence of permanent objects existing outside of the self.
Kant states the thesis of the Refutation clearly enough: ˜˜The mere, but
empirically determined, consciousness of my own existence, proves the
existence of the objects in space outside of me™™ ( ·µ). There is,
however, a certain ambiguity to the notion of ˜˜the mere, but empirically
determined, consciousness of my own existence.™™ It is initially unclear
whether consciousness of one™s existence is empirically determined by
the mere fact that it entails the existence of some indeterminate empiri-
cal representation, or whether empirical determination requires deter-
mination of the position (and hence the content) of that empirical
representation in time.
In the literature, empirically determined consciousness of my own
existence has been rightly understood in this latter, stronger sense.
Kant explicitly denies that the mere consciousness of our own existence
requires an experience of outer objects ( ··). Thus, the kind of
representation of self to be had by transcendental apperception alone is
not su¬cient to provide a premise for the argument. In fact, recent
interpreters have rightly taken this consciousness to be the inner experi-
ence that Kant identi¬es with self-knowledge.µ Kant uses the term
˜˜experience™™ in many places as a synonym for knowledge, in which we
are even said to know the objective temporal relations of what we
perceive (see  ±). And in his own subsidiary remarks to the Refuta-
tion, he identi¬es inner experience with knowledge ( ··).
Once we assume that we have self-knowledge the argument lends
itself to a straightforward reconstruction. Self-knowledge requires
knowledge of a certain temporal order among one™s inner states. Knowl-
edge that inner states have a certain temporal order requires that there
actually be a certain temporal order governing those inner states. And,
assuming that such a temporal order is only determinable with respect
to something that is outside of us, then there must be something that
exists outside of us.
In the argument, Kant ¬rst appeals to a premise from the First
Analogy. All time-determination requires something permanent in per-
ception. We may recall that consciousness of one™s inner states is
°° Kant and the demands of self-consciousness
inherently successive. The replacement of one state of consciousness by
another state of consciousness after it, is not enough to support a
representation of succession. There must be something that is represen-
ted as constant through that transition. It is then through one™s repre-
sentation of this item that persists over one™s representations, while other
contents of representation undergo change, that one becomes conscious
of the transition from one temporal state to another. Even to have a
belief that something has occurred at some point in time, I need
something that I represent as persistent or even permanent relative to
which something can appear to me to be an event.
Initially, it seems that what I take as permanent might merely be
something that is represented by me as permanent, without it, in fact,
being permanent. How can it be successfully argued that my own
existence in time presupposes something permanent that cannot itself be
a mere represented? While Kant assumes in the First Analogy that all
permanent objects are objects that are to be met with by us in space, he
does not attempt to o¬er any proof that this is so in that context. The
proof of this claim that he o¬ers in the Metaphysical Foundations has severe
di¬culties, as we have already seen. But, in order to provide a proof of
the dependence of beliefs about inner states on the existence of external
objects, the Refutation must also argue for the claim that only objects to
be met with outside of me can be permanent.
The permanent object that I need to order my subjective states in time
cannot be purely private, for I would then have no way to distinguish
between a veridical and a non-veridical temporal ordering of my mental
events. If all I had to go on were my own private experiences, then I would
not even have the notion of a temporal perspective that is di¬erent from
the one I am taking and then I would have no basis for thinking that any of
my beliefs about the temporal order of my experiences could ever be false.
However, even if I need some representation of the possibility of a
di¬erent take on my inner experiences than the one I have now in order to
be able to take myself to be making a judgment about my inner states, it is
still not obvious that this cannot be just another inner take of mine with
which I might compare my present experiences. This initially suggests
that regularities in the occurrence of purely inner psychological states
could be su¬cient to allow for the determination of the relations of
co-occurrence and successiveness of states.
Henry Allison argues that consciousness of a succession of representa-
tions is, at the same time, a succession in my consciousness.· But this
view would be rejected by someone like Paul Guyer, who thinks that one
The argument against idealism
can have a momentary consciousness of a succession of states. Now,
regardless of whether one could represent a succession of states at a
moment or not, and regardless of whether one could then represent
something persistent at a moment or not, a representation of something
persistent from the vantage-point of a purely subjective take on things
would give me, at best, an apparently persistent object. The claim that
the persistent or permanent might be a mere representation would
reduce the permanent to an apparent permanent. Such a merely appar-
ent permanent would then yield only an apparent temporal order
requiring in its turn a spatio-temporal permanent which could not be in
turn purely temporal on pain of in¬nite regress. Thus, without some
permanent that is outside of my inherently successive inner experiences,
I would not even be able to make veridical claims about the order of
those inner experiences, and so I would not have inner experience at all.
This seems to be part of the claim that Kant is making in a note added to
the preface to the second edition, in which he argues that representa-
tions themselves ˜˜require as such a permanent distinct from them in
respect to which their replacement-change and hence my existence in
time, in which they replace themselves, can be determined™™ ( ©).
We need not have an uninterrupted perception or representation of a
particular in order for that particular to serve as a basis for providing a
temporal order for our representations. The permanent is supposed to
be something that we can assume to continue to exist during any
interruptions of our conscious experience, such as when we sleep.
Without the notion of something that continues even while our own
train of representations is interrupted, we would have no basis for
accounting for non-conscious periods of our existence within our ex-
[T]he representation of something permanent in existence is not identical with a
permanent representation; for this [representation] can be very changeable and
subject to replacement-change, as all of our representations and even those of
matter [are], and relates itself to something permanent, which must therefore
be a thing that is distinct from all my representations and external [to them],
the existence of which is necessarily contained in the determination of my own
existence and makes up only a single experience with it [the determination of
my own existence] that would not even occur internally, if it were not (in part)
also external. ( ¬©)
Kant admits that even our representations of physical objects are not
themselves enduring representations. But his concession does not mat-
ter, since the persistent or permanent cannot be something that is
° Kant and the demands of self-consciousness
merely represented as persistent. Everything in me exists only insofar as
it can be represented by me. If it is true that the permanent cannot be
something merely represented by me, then it follows that the permanent
cannot be something in me, something internal to consciousness itself.
Kant concludes that the permanent must be something outside of me.
The assumption that the permanent objects outside of us are to be
understood as spatial objects is introduced in the second note to the
Refutation. According to the note, the permanent objects in question
must be material objects. Kant™s observation has some force that we do,
in fact, derive our consciousness of temporal relations from changes in
such external objects as the sun. But solar motion relative to terrestrial
objects is relative motion. Each of these objects undergoes changes, and
none of them can be regarded as permanent from a cosmic perspective.
None of the objects that we perceive are permanent objects in the strict
sense. The fact that the objects that we perceive are not permanent,
forces Kant to admit that we do not derive the permanence of external
objects from external experience ( ·), but presuppose it a priori as a
condition for determining temporal relations. This certainly seems to
put objects in space on a par with objects existing only in time, in other
words, with representational states.±°
The problem of change and hence of permanence also arises for
purely spatial objects. For, even though Kant thinks that physical
properties are only to be regarded as temporal states in virtue of their
relation to mental events, such physical properties do present them-
selves to us as temporal states of physical objects. However, in the case of
physical objects and their states, it is possible for us to give a tenseless
characterization of those objects, while we cannot provide any such
account of mental states from the standpoint of inner experience. We
can assign a determinate tenseless temporal order to our inner states,
but we are able to do this only by appeal to the existence of physical
objects that are not essentially tensed in the same way that mental states
Now it is important to note at this stage that external objects are
privileged in the determination of the temporal order of events (includ-
ing mental events) because they are public objects. Failure to appreciate
the fact that external objects must be understood as public objects has
seriously undermined some discussions of the Refutation. Thus, C. D.
Broad argues that the external objects together with the spaces to which
they belong which we perceive are something private and dependent on
the mind of the individual percipient.±± But it is only once one realizes
The argument against idealism
that external objects cannot be private objects precisely because they are
external objects that the signi¬cance of external objects in providing an
independent standard for determining the temporal order of our beliefs
becomes apparent.Whatever external objects may turn out to be, it is
their very externality from the contingent order of succession of states in
my or your consciousness that makes them an appropriate basis for
assigning a temporal order to my or your states. For it is their very
externality to my or your consciousness that makes them capable of
providing independent con¬rmation or discon¬rmation of the temporal
order that each of us subjectively assigns to our individual states and
thus gives sense to such a subjective order in the ¬rst place.

    ¬   ¦ ©   ¤ ©    © ®      § µ  ®
The Refutation of Idealism is, in many respects, a revision of the
argument articulated against skepticism about the external world in the
Fourth Paralogism in the ¬rst edition of the Critique. There, Kant also
attempts to show that we have an immediate experience of the external
world, as he does in the Refutation: ˜˜that is, the consciousness of my
own existence is also an immediate consciousness of the existence of
other things outside of me™™ ( ·). But, after this, the argument diverges
substantially. For in the ¬rst edition Fourth Paralogism, Kant is content
to argue for this immediacy on the basis of the fact that external objects
are objects represented by me immediately in my outer sense, just as my
inner states are immediately represented to me through my inner sense
( ·±). In the new argument in the Refutation, by contrast, my con-
sciousness of my inner states turns out to be mediated by a consciousness
of objects outside of me. And these objects outside of me are precisely
not mere representeds, as the argument in the ¬rst edition had main-
tained.± The argument attempts to establish that one™s consciousness of
one™s own existence is coupled to an immediate consciousness of things
which are distinct from it.±
Although it is clear that Kant wants to claim that outer objects are
supposed to be experienced directly, it is less clear if one has any direct
consciousness of the objects of one™s ˜˜inner time-consciousness.™™ The
evidence on this issue is somewhat ambiguous. In private notes devoted
to rethinking his argument, Kant claims that ˜˜empirical consciousness
of myself, which constitutes inner sense by no means occurs immediate-
ly™™ (Re¬‚ection µµ, Ak. ©©©, p. °). But this does not really settle the
question of whether one has immediate knowledge of the objects of
° Kant and the demands of self-consciousness
inner experience. The process of introspection may in some sense be
mediated by its relation to external objects while still providing one with
non-inferential knowledge of one™s inner states. This possibility seems to
have been overlooked by Guyer, who cites the Re¬‚ection as evidence
against the immediacy of inner experience.± In the Refutation, Kant
states that ˜˜inner experience is itself only mediate and possible through
external™™ experience ( ··). But there is no evidence that he wants to
suggest that our knowledge of inner experience is therefore purely
inferential, as Guyer suggests that he does.
It would appear that, for Kant, the non-immediacy characteristic of
my empirical self-consciousness is a function of its mediation by con-
sciousness of objects outside of me. The claim that all inner experience
is parasitic on outer experience might be taken to mean that all inner
experience is also at the same time an experience of something outer,
or it might mean something weaker. It might, for instance, mean only
that inner experience somehow presupposes outer experience.±µ Kant
needs the stronger claim that any determinate consciousness or belief
about something temporal must involve a consciousness of something
spatial. This is the full import of his thesis that ˜˜the representations of
external senses make up the actual content of inner sense with which we
occupy the mind™™ ( ·). On the other hand, it is also true that any
consciousness of something outside of me is also a consciousness of
something that is present to my consciousness and hence a part of my
inner experience.
Kant would hardly deny that our experience of objects in space is
itself mediated by consciousness of inner states. For according to him,
objects in space are only in time in virtue of their relation to our inner
states. Without the mediation of our time-consciousness one could not
even experience objects as persistent through changes in their states.
There would not be any changes in state there to be experienced in the
¬rst place. At best, Kant can sustain the claim that while our conscious-
ness of inner and outer objects are mutually dependent, such mutual
dependence does not preclude immediacy of awareness.
The asymmetry with respect to the immediacy claim must be be-
tween the way judgments about inner states and about outer states is to
be justi¬ed. As knowledge that is expressible in a judgment, our knowl-
edge of the external world is propositional, but it is also the non-
inferential articulation of what we immediately perceive and hence
involves direct apprehension. Immediate knowledge of the external
The argument against idealism
world thus has features of both immediate propositional knowledge and
of direct apprehension. Now judgments about inner states can only be
justi¬ed by appeal to knowledge of outer states. By contrast, judgments
about outer states could be justi¬ed even if we did not have any
self-knowledge at all, although it would be impossible to have any
knowledge of outer states without some consciousness of our inner
states. For outer states are only (successive) temporal states in virtue of
their relation to inner states.
The claim that all knowledge of inner states is parasitic on knowledge
of outer states has some strong support in Kant™s position. Remember
that the application of categories is supposed to make objective claims
about spatio-temporal events possible. Now there is strong evidence that
Kant restricts the direct application of the categories to the spatial
objects of outer sense ( ±¬.).± This restriction requires that all mental
events have corresponding physical states. While Kant does not seem to
believe that mental events must have corresponding physical states, he
does believe that we can have no experience of mental states that do not
have corresponding physical states. The very conditions governing
ascription of mental states collapse in the case of disembodied souls. We
cannot really make sense of the representational states involved in the
praxis of disembodied souls, since the conditions governing our ascrip-
tions of content do not hold with respect to them. We do not have any
self-knowledge at all which does not also include knowledge of objects in
Despite the emphasis Kant puts on immediacy of representation in
the Refutation, Guyer has argued that the proof for ˜˜the immediate
consciousness of the existence of external things™™ and the proof ˜˜that
external experience is immediate™™ ( ·“··n) cannot be taken serious-
ly. This is supposedly because Kant does not appeal to any premise
involving immediacy in his argument.± Kant does claim that the im-
mediate consciousness of things outside us is not assumed, but proven (
··n). Guyer suggests that there is no premise entailing immediacy
available to Kant.± In part, Guyer™s claim is based on Kant™s alleged
commitment to a premise that is undeniably to be found in the ¬rst
edition, that one only has immediate consciousness of one™s representa-
tions.° This thesis may be found in the Fourth Paralogism in the ¬rst
edition ( ·±). However, Guyer o¬ers no argument to show that Kant
retained commitment to this particular immediacy thesis in the second
edition. The Refutation of Idealism seems, rather, to be based on
° Kant and the demands of self-consciousness
rejecting the restriction of immediate awareness to representings that
formed the cornerstone of the argument against idealism in the ¬rst
edition of the Fourth Paralogism.
The natural reading of the immediacy claim in the Refutation is that
our knowledge of external objects (rather than knowledge of our repre-
sentings) must be immediate if it is to justify claims of self-knowledge.
Guyer maintains that this cannot be the case because claims to knowl-
edge of external objects will be justi¬ed in many cases by appeal to
putative instances of self-knowledge.± Now it is true that putative
instances of self-knowledge could provide inductive support for some
knowledge claim about external objects. But this self-knowledge will
itself always presuppose some knowledge of external objects. Apart from
such potential dependence of certain knowledge claims about external
objects on claims to self-knowledge, there does not seem to be any
reason to attribute to Kant the view that self-knowledge per se ever
provides a basis for con¬rming or discon¬rming knowledge claims
about external objects.
Guyer asserts that only if claims concerning self-knowledge and
knowledge of outer objects are on a par can the argument of the
Refutation be saved from circularity. Causal knowledge must be derived
from induction on subjective successions while knowledge of subjective
succession depends on knowledge of causal laws. The worry of circu-
larity that Guyer raises against Kant disappears once one recognizes the
possibility of direct awareness of subjective succession. And so it does
not seem necessary to go to the expedient of appealing to our knowledge
of external causes to account for our being able to justify our beliefs
about our inner states. But, given Guyer™s interpretation of time-con-
sciousness, it is not surprising that the argument which Guyer views as
an appropriate substitute for an argument from an immediate con-
sciousness of things outside of me is based on inferential causal knowl-
edge of objects outside of us. This is, however, a singularly unpromis-
ing line of attack. In giving up the immediacy premise, Guyer winds up
attributing to Kant and defending a version of precisely the kind of
idealism Kant o¬cially sets out to refute! In diagnosing the position of
˜˜problematic idealism,™™ that he wishes to reject, Kant notes that the
˜˜problematic idealist™™ assumed that:

the only immediate experience is an inner one, and that one merely infers the
existence of external things from it, and that, as in general, when one infers
determinate [speci¬c] causes from given e¬ects, [this is] undependable, because
The argument against idealism
the cause of the representations, which we perhaps falsely attribute to external
things, may lie in us ourselves. ( ·)

Any judgment to the e¬ect that a certain causal relationship holds
between the objects of our beliefs and those beliefs themselves seems to be
open to skeptical attack. If all we know are the e¬ects of a certain causal
connection, our belief that those e¬ects are the result of a certain kind of
cause is itself always open to skeptical doubt. A di¬erent set of causes than
the ones which we hold to be the source of our beliefs might in fact turn out
to be their true source. Something we believe to have a cause outside of us
may turn out to have a cause inside of us. The criticism of causal
explanation as a basis for belief in the external world is also echoed in later
Re¬‚ections, as well as in the Fourth Paralogism of the ¬rst edition
(Re¬‚ection µµ, Ak. ©©©, p. ±;  ). So this is one of the few points
that remain unchanged in Kant™s arguments against idealism.
The Refutation does not rule out the possibility that one has a direct
experience of objects outside of one based on one™s causal relations to
those objects. Kant™s worry about skeptical doubt concerns the postula-
tion of particular causes as explananda for our representations. On the
other hand, he is committed to the Humean idea that causal connection is
imperceptible ( ±/ ±··). The imperceptible character of causal
connection makes it di¬cult to see how we could have immediate
(non-inferential) knowledge of causal relations in any sense. And Guyer
does interpret the fact that ˜we can only know our existence in time in
commercio™ (Re¬‚ection ±±, Ak. ©©©, p. ±) to mean ˜˜that we must
perceive our bodies in causal interaction with the other bodies we use for
time-determination.™™ It is rather odd to base the argument of the
Refutation on a premise that Kant not only never states but also explicitly
denies in the Schematism which is supposed to lay the groundwork for his
theory of time-determination: ˜˜no one will say that this [category], for
instance causality, can be intuited through the senses and is contained in
appearance™™ ( ±··/ ±). Another, I think more plausible, reading of
Kant™s claim, is that our inner experiences must be causally connected to
our bodies. And our bodies are in turn, of course, causally connected to
other bodies. Causal interaction immediately links our representations
and objects outside of us, but the causal link in question is not something
that we perceive. We have an immediate perception of objects outside of
us because we interact causally with them. But our knowledge of the
causal relation is inferential and part of a global systematization of nature,
whereas our knowledge of the perceptual object is immediate.
° Kant and the demands of self-consciousness

     ¦ µ   © ® ® ¤   © ® §  µ  © ¤  ¦ µ  © ®  
 ®  ® ¤® ¬  ® 
Eckart Forster maintains that, by arguing that the determination of my
existence in time is possible only through a thing outside me and not
through the mere representation of a thing outside me, Kant refutes not
only Cartesian idealism, but also his own transcendental idealism, since
the latter requires ˜˜that what we call outer objects are nothing but mere
representations of our sensibility™™ ( °/ µ, my italics). Forster concludes
that either the Refutation of Idealism refutes Kant™s own transcendental
idealism, or it fails to refute idealism at all.µ The objection fails to take
account of the two di¬erent senses of the expression ˜˜external™™ that
Kant works with. In the Fourth Paralogism, Kant notes that ˜˜the
expression outside of us carries with it an unavoidable ambiguity, in that it
sometimes means something that exists distinguished [by us] from us as
a thing in itself and sometimes means that which belongs to external
appearance™™ ( ·). The ambiguity in the notion of externality carries
over to the notion of internality. The distinction between appearances
and things in themselves itself turns out to be ambiguous in a manner
that is analogous to the ambiguity in the inner“outer distinction. For we
can draw a distinction between appearances and things in themselves
within experience. In this case the distinction between appearances and
things in themselves corresponds to the distinction between what is
inner to experience, in the sense of what is part of an individual point of
view, and what is outer, in the sense of what is external to an individual
point of view. But we can also draw a distinction between appearances
and things in themselves in which anything belonging to our experience
(including what is external to individual points of view) is an appear-
ance, and anything which is completely external to experience is a thing
as it exists in itself.
Thus, both the inner“outer distinction and the distinction between
appearances and things in themselves are ambiguous. There is a distinc-
tion to be drawn within experience between the private and the public,
and a distinction between what is public, but nevertheless dependent on
the way we must together experience the world, and the way the world is
independently of being experienced by us. The distinction between
appearances and things in themselves within experience is an empirical
distinction, while the transcendental distinction concerns the relation of
experience to what is outside of experience itself.
Kant™s response to Cartesian idealism in the Refutation is based on
The argument against idealism
an empirical understanding of ˜˜external.™™ Transcendental idealism is
committed to denying that external objects as represented in the empiri-
cal sense are external to us in the transcendental sense. The context of 
°/ µ in which he stresses his thesis that what is external to us in the
spatial sense is not as such a thing in itself indicates that he is concerned
with what is ˜˜outer™™ in the empirical sense.
In fairness to Forster, an argument against idealism in the empirical
sense does not address the question of whether the external world might
not turn out to be the representation of an individual mind after all,
once we move to the level of transcendental re¬‚ection. But, on the
other hand, the thesis that consciousness of oneself in time presupposes
the existence of things in themselves existing outside of that self-con-
sciousness in a transcendental sense, is not only compatible with Kant™s
transcendental idealism, it can even make a distinctive contribution to
the ultimate success of his argument against skeptical idealism. For,
without a defense of the existence of radically mind-independent things
in themselves, the problem posed by skepticism seems to be merely
pushed back a stage.
Some commentators have inferred that objects outside of us must be
things in themselves because they are not objects existing purely in my
private space. This is clearly the wrong reason for inferring that they
must be things in themselves.· Kant™s thesis that objects in space are
transcendentally ideal, and not transcendentally real as things in them-
selves would have to be, means that, at the very least, the same object
cannot be transcendentally ideal and transcendentally real under the
same description. But it need not mean that the object cannot exist both
independently of us in the spatial sense and in a much more radical
sense. Under a spatial description, it may turn out to be dependent on
our subjectivity, while it may turn out to be completely independent of
our subjectivity under some other description.
The Refutation of Idealism makes no explicit reference to things as
they exist in themselves, and the passage at  ¬© that I have already
quoted makes it quite clear that Kant thinks of the objects outside of us
that are required for inner experience as themselves belonging to
experience. They cannot be things in themselves in the transcendental
sense of the term. However, in later re¬‚ections concerning the problem-
atic of the Refutation, Kant does try to argue that inner experience is
tied to the existence of things in themselves in the transcendental sense.
These are thoughts he put to paper in response to queries by Kiesewet-
ter. In meditations dating from the ±·°s, he argues that the transcen-
±° Kant and the demands of self-consciousness
dental ideality of objects provides the basis for a refutation of skepticism
concerning the existence of things in themselves. To the extent that he
can make this claim stick, he has a compelling response to those who
claim that he has simply given the game of responding to skepticism
away by restricting his knowledge claims to appearances.
The idea that our knowledge of the independent existence of ap-
pearances should somehow be linked to some consciousness of things as
they exist in themselves independently of the conditions governing our
knowledge should not be a surprising one when viewed from the
perspective of a dual-aspect interpretation of transcendental idealism.
According to the dual-aspect interpretation, appearances and things in
themselves are two di¬erent descriptions of the same things. To describe
something as it is in itself is, of course, to describe it in more fundamental
terms than the way it must appear to us spatio-temporally, but these are
nevertheless two distinct ways of describing something. In order to
provide an adequate response to skepticism, a refutation of idealism
must at least prove that I have knowledge that things exist in themselves
outside of my mind. Now the Refutation is successful at making good on
this demand at the empirical level. Objects outside of me in space are
genuinely outside of my mind, as my mind must appear to me. How-
ever, it might still be argued that there could be no objects that are
independent of my individual mind, as it is in itself.
In the ±·°s, Kant tries to address this worry, by showing that we do
have a kind of knowledge that there are radically mind-independent
objects, although we cannot know how we should describe such objects.
This does not violate critical strictures on knowledge of things as they
are in themselves. Indeed, Kant regards knowledge that things in
themselves exist as transcendental knowledge. It is not knowledge of the
nature of those things as they exist in themselves. Such knowledge would
be transcendent knowledge for him. The notion of a thing in itself must
be intelligible, and hence representable by us in some very minimal and
negative sense if we are to make sense of the distinction between
appearances and things in themselves. Otherwise this is a distinction
without a di¬erence, and transcendental idealism collapses into tran-
scendental realism.
In general, Kant does not commit himself to the non-representable
character of things in themselves, but rather to their being unknowable
according to the canons of theoretical knowledge. This leaves room for
an immediate relation to things in themselves so long as this awareness is
not interpreted as full-blown theoretical knowledge. Such knowledge
The argument against idealism
that there are things in themselves cannot entail substantive knowledge
of the nature of those things in themselves. It does, however, commit
one to a form of negative theology; one knows, for instance, that things
in themselves cannot be spatio-temporal objects when conceived in
themselves. To a¬rm the impossibility of treating objects in space as
available under some non-spatio-temporal description is just to deny the
legitimacy of the double-aspect approach to transcendental idealism in
favor of some form of double-object theory. Even some interpreters who
have advocated a double-object approach have also argued that the
Refutation requires reference to things in themselves.
Kant maintains that representations of objects in space are only
possible to the extent that there is some thing in itself to which we are in
a ˜˜real™™ relation:

But if it is shown that the determination of our own existence in time presup-
poses the representation of space in order even to represent to oneself the
relation of the determinations of inner intuition to a constant object :zum
bleibenden Objekt9 . . . then external objects can be secured a reality (as things in
themselves) precisely through one™s not taking their intuition as that of a thing
in itself; for if it [the intuition of space] were of such [a thing in itself ] and the
form of intuition were the form of a thing, (which) were to belong to it [that
thing] in itself even without the particular character of our subject, then it
would be possible that we would have the representation of such a thing
without that thing existing. However there is a particular kind of intuition in us
which cannot represent what is in us, hence what exists in the change of time
:Zeitwechsel9, since it would then be thinkable in the mere representation of
temporal relations, therefore such a representation must subsist in the real
relation :wirklicher Beziehung9 to an object outside of us and space really
means something which in being represented in this form of intuition is only
possible through the relation to a real thing outside of us. “ Therefore Refuta-
tion of skepticism. Idealism. Spinozism. also of materialism, predetermin-
ism. (Re¬‚ection ±·, Ak. ©©©, pp. ·“)

We cannot have a representation of a spatial object without that object
somehow existing outside of our spatial intuition. The thing in itself is
represented by us as a real object in space, although, in fact, it is also in a
real relation to the subject that cannot be spatio-temporal in character.
The suggestion is that metaphysical or transcendental realism is false.
Metaphysical realism insists that our beliefs about the world may be
completely false, since the objects of our beliefs are completely indepen-
dent of our beliefs. The complete independence of the world from our
beliefs opens up the prospect of skepticism. There is no way for us
± Kant and the demands of self-consciousness
decisively to determine from the nature of our beliefs whether they are
radically misguided or not.
The re¬‚ection o¬ers some indications as to how skepticism about the
external world may be refuted. We must reject the assumption that we
could have spatial representations of objects divorced from real relations
to objects existing outside of our subjectivity. Kant does not leave it at
this petitio principii. He also attempts to o¬er a vestige of an argument for
the rejection of this assumption. The argument turns on the distinction
between the representation of a spatial and the representation of a
temporal relation. The key premise of the argument is that a spatial
representation cannot be a representation of what is in us or of the
passage of time, ˜˜Zeitwechsel.™™ This appears to be another petitio. We
want precisely to establish that we have a representation of something
that is truly outside of us. We need to demonstrate that space is not
merely a representation of inner experience. If space turns out to be a
representation of inner experience and inner experience is essentially
temporal, then our representation of objects in space would also involve
a representation of a passage of time.
Fortunately, Kant o¬ers a sketch of how the distinction between the
faculty of representing things temporally and the faculty of representing
things spatially might be justi¬ed. The argument would have to rest on a
proof of the conditional with which he begins the re¬‚ection. One would
have to show that temporal relations can only be established by appeal
to some persistent object in space that is not essentially temporal.
Objects given to us a priori through our temporal representation are by
de¬nition essentially temporal. One must show that consciousness of
temporal relations is parasitic on consciousness of objects in space that
are not essentially temporal. Then one may argue with some conviction
that it is only in virtue of a mode of givenness of objects that is
independent of the essential temporality of inner experience that we can
experience temporal relations at all. Our consciousness of objects of
inner sense, and indeed of inner sense itself, will turn out to be parasitic
on the existence of the spatial objects of outer sense. Kant insists that
both an intellectual consciousness of things outside of me and a determi-
nation of those things in space must co-occur with the determination of
my existence in time. In making this claim, he links empirical self-
consciousness to an immediate consciousness of things outside of one in
both the transcendental and empirical senses.
Emphasis on the immediate givenness of one™s existence through a
purely intellectual self-consciousness leads Kant to fall back on his old
The argument against idealism
term for intellectual self-consciousness. In Re¬‚ection µµ, the con-
sciousness of things outside of me is referred to as an ˜˜intellectual
intuition,™™ where this intellectual intuition provides one with ˜˜no knowl-
edge of things.™™ The immediacy of existence provided by the act of
self-consciousness had suggested the notion of an ˜˜intellectual intuition™™
to him in the seventies. Kant now endeavors to link this immediacy of
existence not only to an event in time, but also to the immediate
consciousness of something in space outside of me. He insists however
that from this immediate consciousness ˜˜I [know] my own empirically
determined existence no more than that of things outside of me (which,
what they are in themselves I do not know)™™ (Re¬‚ection µµ, Ak. ©©©,
p. °).
Kant maintains that the immediacy in our consciousness of external
things is the basis upon which such an interpretation must build. But,
according to Guyer, the mere occurrence of intuitions with either temporal
properties or even the phenomenological form of spatiality is not itself
su¬cient to provide even a representation of the temporal relations of
these representations, so that questions of immediacy or mediation arise
only once we have interpreted spatial and temporal experience.° It is, of
course, true that the discussion of the status of consciousness as immediate
or inferential is a topic of more abstract re¬‚ection, but this is not because
representations only represent when they are interpreted or judged. For
Kant clearly does not think that representations need to involve inference
in order to be represented at all.± In fact, he pushes a point that is quite
di¬erent from the point that Guyer derives from the passage.
Kant argues that we have an immediate and, hence, non-inferential
consciousness of ourselves as passive beings, as beings to whom data is
merely given. This is his justi¬cation for the claim that space cannot be
within me. The assumption is that anything in me is a function of my
spontaneity. Without an immediate consciousness of oneself as a passive
being, one could not even represent things as being outside of oneself.
Although the concepts of passivity/receptivity and activity/spontaneity
are highly charged technical terms of philosophy, he suggests that they
serve to articulate an immediate experience of the world and, indeed, the
content of one™s own experience as not completely of one™s own making. If
one is a passive being, then there must be things which exist outside of one,
because otherwise one could not even have an inner experience.

The intuition of a thing as outside of me presupposes a determinability of my
subject in which I am not myself the determinator, which therefore does not
± Kant and the demands of self-consciousness
belong to spontaneity, because the determinator is not in me. And in fact I
cannot think any space as in me. Therefore the possibility of representing things
in space in intuition is grounded on the consciousness of a determination
through other things, which [consciousness] means nothing other than my
Original passivity in which I am not active. That dreams bring about deception
concerning existences outside of me does not demonstrate anything against
this; for external perceptions would have had to precede them at any rate. It is
impossible to get a representation from something outside of me without being
in fact passive. (Re¬‚ection µµ, Ak. ©©©, p. °·)

The refutation of idealism ultimately depends on the idea that con-
sciousness of a mere subject of thought is not enough for representing
anything as genuinely in space and time and, hence, for representing
anything as distinct from space and time themselves. One must be able
to represent some particular as distinct from one™s own subjectivity, if
one is even to represent oneself as a numerically identical particular in
time. Without some other individual that is external to one, one does not
yet have a grip on the numerical distinctness of one™s own identity as a
particular individual.
The link between intelligibility for us and the content of beliefs also
provides some support for a rejection of global skepticism. It undercuts
the obvious objection that we may have concepts and quasi-perceptual
experiences (imaginings) of many things that do not actually exist. It
cannot be that everything that appears to me in perception is actually
there, since there are delusive perceptions. The best that can be hoped
for is an argument against global error together with a framework for
tracking down local error. Our capacity to identify and reidentify
particulars across di¬erent spaces and times, together with our ability to
classify them into sorts, goes a long way to dealing with the problem of
local error. Global error would undermine the conditions under which
we can give representations whatever content they have.
Without the premise that global error is impossible, Kant has no
serious prospects for blocking the claims of subjective idealism. He
argues for the impossibility of global error by appeal to a version of the
principle of charity. In order to take some claim as a judgment one must
be able to make sense of that claim. However, one can only make sense
of a claim against the background of certain other beliefs that one can
also take to be true.
The appeal to some form of the principle of charity gives rise to
another form of idealism however. But the idealism in question is a
conceptual idealism. The governing idea of this conceptual idealism is
The argument against idealism
that we cannot make sense of the notion of a world that does not
conform to the conditions under which we can apply concepts at all.
Even insistence on the existence of things in themselves does not violate
this principle. Kant thinks that we can conceive of things in themselves
only as objects of an understanding stripped of the conditions governing
sensibility. This is why he often identi¬es things in themselves with
noumena (intelligible objects). Since Kant acknowledges the possibility
of merely imagining the existence of particular external objects “ this
possibility is the topic of the third and ¬nal note to the Refutation “ he
must have some premise that blocks the move from local skepticism
about external objects to global skepticism about their existence. The
intelligibility premise also undercuts skepticism about other minds, since
it calls the skeptic™s assumption that we could be completely wrong in
our ascription of minds to other beings into question.
If self-a¬ection, that is the determination of the content of one™s
experience, were possible completely independently of one™s environ-
ment, then the empirical idealist™s position would be unassailable. Since
inner sense involves self-a¬ection, saying that self-a¬ection is possible
without a¬ection by external objects is another way of saying that an
inner sense is possible without an outer sense. Kant tries to rule out such
a possibility by arguing that it is incompatible with the existence of
consciousness of oneself as a spontaneous being:

If we were only a¬ected by ourselves but without noticing this spontaneity, then
only the form of time would be found in our intuition: and we would not be able
to represent any space to ourselves (an existence outside of us). Empirical
consciousness as a determination of my existence in time would therefore go
around in a circle and presuppose itself “ above all it would be impossible, since
the representation of something persistent would be missing, in which there is
no continuous synthesis as in time. (Re¬‚ection µµ, Ak. ©©©, p. °)

For those who ¬nd the idea that we are spontaneous beings implausible
to begin with, this does not have much bite. In somewhat more general
terms, Kant argues that we cannot make sense of ourselves as inter-
preters of experience without the idea that there is something in experi-
ence that is not up to us. We can only have the representation of our
activity of interpreting, if there is something of which we are also
immediately conscious that we must represent as distinct from this
activity. This is why Kant infers the existence of an immediate con-
sciousness of passivity or receptivity from the existence of an immediate
consciousness of our spontaneity: ˜˜the concept of the mere passivity in a
± Kant and the demands of self-consciousness
state of representation . . . is not inferred, since we do not perceive in us
the cause of the existence of a perception, yet it is an immediate
perception™™ (Ak. ©©©, pp. °·“°). The immediate consciousness of
our spontaneity is something we can only be aware of if we have an
equally basic consciousness of our receptivity. Once we have a con-
sciousness of our spontaneity we are then in a position to know that
experience of objects in time as a form of inner sense must be connected
to an immediate experience of objects in space. Now it is not altogether
implausible to argue that we have an immediate experience of things
not being up to us, but also of things needing interpretation by us. This
could indeed be argued to be constitutive of our self-consciousness as
¬nite rational beings.
While Kant could have argued for the existence of a fundamental
distinction between what is internal to an individual point of view and
what is external to all of our points of view had he not insisted on a sharp
distinction between the conditions under which objects are intelligible
to us and the conditions under which they are intelligible tout court, it is
signi¬cant that in the end he wants to argue that our very consciousness
of ourselves as experiencers depends on the existence of objects that are
not only publicly and spatially accessible to us, but which are also
radically independent of us.
   °   ± ±

Empirical realism and transcendental idealism

In the last chapter, I explored Kant™s argument that we can only make
sense of claims to self-knowledge if we commit ourselves to the existence
of objects that exist outside of us. Kant identi¬es this realism with
respect to objects that are experienced as outside of us in experience
with empirical realism. More controversially, he argues that the kind of
knowledge of external objects that empirical realism requires, can only
be established by appeal to his doctrine of transcendental idealism. It is


. 6
( 9)