. 1
( 9)


«  ®   ® ¤    ¤    ® ¤   ¦   ¬ ¦-   ®   ©  µ  ®   

In Kant and the demands of self-consciousness, Pierre Keller examines
Kant™s theory of self-consciousness and argues that it succeeds in
explaining how both subjective and objective experience are poss-
ible. Previous interpretations of Kant™s theory have held that he
treats all self-consciousness as knowledge of objective states of
a¬airs, and also, often, that self-consciousness can be interpreted as
knowledge of personal identity. By contrast, Keller argues for a
new understanding of Kant™s conception of self-consciousness as
the capacity to abstract not only from what one happens to be
experiencing, but also from one™s own personal identity. By devel-
oping this new interpretation, Keller is able to argue that transcen-
dental self-consciousness underwrites a general theory of objectiv-
ity and subjectivity at the same time.

° ©     «  ¬ ¬   is Associate Professor of Philosophy at the Uni-
versity of California, Riverside. He has published a number of
articles on Leibniz, Kant, Hegel, Heidegger, and Husserl.

University of California, Riverside
°µ¬©¤   ° ®¤© ¦  µ®©© ¦ ©¤§
The Pitt Building, Trumpington Street, Cambridge, United Kingdom

©¤§ µ®©© °
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
Ruiz de Alarcón 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa


© Pierre Keller 2004

First published in printed format 1999

ISBN 0-511-04009-1 eBook (netLibrary)
ISBN 0-521-63077-0 hardback
ISBN 0-521-00469-1 paperback

Acknowledgments page vi

± Introduction ±
 ±
Introducing apperception
 µ
Concepts, laws, and the recognition of objects
 Self-consciousness and the demands of judgment in the
µ Self-consciousness and the unity of intuition: completing
the B-Deduction
 Time-consciousness in the Analogies ±±
· Causal laws ±±
 Self-consciousness and the pseudo-discipline of
transcendental psychology
 ±
How independent is the self from its body?
±° The argument against idealism ±
±± Empirical realism and transcendental idealism ±·



The contents of this book have germinated in a long process going back
to my ¬rst Kant seminar with Dieter Henrich in Heidelberg. Although I
sometimes criticize his views, his in¬‚uence is obvious in my work. I am
also strongly indebted to discussions with Georg Picht, Enno Rudolph,
Harald Pilot, and, especially, Ru
¨diger Bittner, dating back to my under-
graduate days in Heidelberg. As a student at Columbia, I was lucky to
be able to take on a new set of intellectual debts. After going to
Columbia to work with Charles Parsons, whose in¬‚uence on my work
on Kant is also patent in this book, I was fortunate to ¬nd Charles
Larmore, Thomas Pogge, Sydney Morgenbesser, and, somewhat later,
Raymond Geuss willing and very challenging participants in discussions
about Kant™s philosophy. I owe a particular debt to Charles Parsons,
Charles Larmore, and especially Raymond Geuss for their helpful
comments on various drafts of this book. Without Raymond Geuss™s
constant criticism, encouragement, and prodding, I am certain that this
book would never have appeared at all. I owe an almost comparable
debt of gratitude to my editor, Hilary Gaskin, who has helped me to see
where the book could be improved and kept at me ¬nally to complete it.
I also wish to thank Gillian Maude for patient help with the copy
Among my colleagues at the University of California at Riverside, I
would be remiss if I did not mention Andrews Reath, David Glidden,
Bernd Magnus, and Larry Wright, each of whom was generous in his
critical comments on my work, in his support, and in his willingness to
engage with Kant™s thought. Fred Neuhouser, Steve Yalowitz, and
David Weberman have also provided much helpful input, as have the
members of the Southern California Kant group, Ed McCann, Patricia
Kitcher, Jill Buroker, Martin Schwab, and Michelle Greer. I am es-
pecially indebted to many discussions with Henry Allison, who has
undoubtedly in¬‚uenced me more strongly than some of my criticisms of
Acknowledgments vii
his views might suggest. My new colleague Allen Wood™s helpful criti-
cisms have led me to make a number of signi¬cant changes. My students
have also led me to rethink a number of things. I wish especially to
mention John Fischer and Laura Bruce, who also helped me with the
proofs and the index.
Finally, I want to thank my parents who instilled an early respect for
Kant and love of philosophy in me, and my brother, Gregory, and
sisters, Karen and Catherine, for having been so supportive of my
projects over the years. My philosophical discussions with Catherine
have also undoubtedly had an impact on the present work. My greatest
debt is to my wife Edith, who has provided me with invaluable criticism
of every draft, and much-needed intellectual and emotional support.
° ±


In the Critique of Pure Reason (henceforth Critique), Kant draws a famous
but elusive distinction between transcendental and empirical appercep-
tion. He interprets the distinction between transcendental and empirical
apperception as a distinction between transcendental and empirical self-
consciousness. He argues that empirical self-consciousness is parasitic
on transcendental self-consciousness, and that any empirical conscious-
ness that has any cognitive relevance for us depends for its cognitive
content on its potential relation to transcendental self-consciousness.
These are strong, but, I want to argue, defensible claims once one
understands the nature of transcendental self-consciousness, as it is
understood by Kant.
The central aim of this book is to provide a new understanding of the
notion of transcendental self-consciousness and show its implications for
an understanding of experience. I develop and defend Kant™s central
thesis that self-consciousness puts demands on experience that make it
possible for us to integrate our various experiences into a single compre-
hensive, objective, spatio-temporal point of view. My interpretation of
his conception of self-consciousness as the capacity to abstract not only
from what one happens to be experiencing, but also from one™s own
personal identity, while giving content to whatever one represents,
shows how transcendental self-consciousness underwrites a general the-
ory of objectivity and subjectivity at the same time.
The leading interpretations seem to be in broad agreement that Kant™s
notion of transcendental apperception is largely a disappointing failure.
Perhaps the dominant tendency has been to dismiss his notion of
transcendental self-consciousness as at best implausible and at worst
incoherent. But even those interpreters who have been sympathetic to the
notion of transcendental self-consciousness have endeavored to give it an
anodyne interpretation that renders it largely irrelevant to a defense of
objectivity or even subjectivity. By simply identifying transcendental
 Kant and the demands of self-consciousness
self-consciousness with objective experience, those interpreters deprive
transcendental self-consciousness of any substantive role in justifying the
claim that our experience is at least sometimes objective, and make it
di¬cult to understand how it could sometimes be merely subjective.
It is not surprising that interpreters have had their problems with
transcendental self-consciousness, despite the fact that it is undeniably a
central notion in Kant™s philosophy. Part of the problem is that Kant™s
notion of transcendental self-consciousness requires a subject of self-
consciousness that is somehow distinct from any subject that we can
experience. The only kind of subject that we seem to be acquainted with
in any sense is a subject that we can experience, an empirical subject,
and so the notion of a non-empirical subject that we could become
conscious of seems to be based on an illegitimate abstraction from actual
experience.± And, even if one concedes that it might be possible to be
conscious of a non-empirical subject of experience, it seems that the only
way we have of making sense of such a subject is by thinking of it as a
mere abstraction from actual experience, in which case it is di¬cult to
see how it could support any substantive claims about what the nature of
experience must be.
Skepticism about whether it is possible to be conscious of a subject of
thought that is somehow distinguishable from the kind of subject that is
knowable through experience leads interpreters to look to consciousness
of personal identity as the only kind of consciousness of self that we
have. Commentators who have resisted the tendency to collapse tran-
scendental self-consciousness into consciousness of personal identity
have often gone to the other extreme of treating all self-consciousness as
a consciousness of judgments that are objectively valid, thus denying
that transcendental self-consciousness is a necessary condition for con-
sciousness of one™s subjective point of view. And even those commenta-
tors who have tried to conceive of transcendental self-consciousness as a
necessary condition of empirical self-consciousness have not had much
to say about how transcendental self-consciousness could be involved in
empirical self-consciousness.
I claim that Kant™s notion of transcendental self-consciousness is
more robust than it has generally been thought to be, but also more
commonsensical than most commentators have allowed it to be. I argue
that the key to a proper understanding of the thesis that our experience
is subject to the demands of self-consciousness is a proper understanding
of the fundamentally impersonal character of our representation of self.
We have an impersonal or transpersonal representation of self which is

expressed in our use of the expression ˜˜I™™ to refer to ourselves. When
each of us refers to him- or herself by means of the expression ˜˜I,™™ each
of us refers to him- or herself in a way that could, in principle, apply to
any one of us. This is the basic, minimal, idea that Kant tries to express
with his notion of transcendental self-consciousness.
I attribute to Kant and defend several further claims about transcen-
dental self-consciousness that are very controversial. I claim that empiri-
cal or personal self-consciousness is parasitic on transcendental or im-
personal self-consciousness. I argue that this amounts to the claim that
we are only able to grasp our own individual identity by contrast with
other possible lives that we might have led. Then I argue that our very
ability to form concepts in general is based on our capacity for transcen-
dental self-consciousness. This capacity for concept formation and use is
displayed in judgments and inferences that themselves depend on our
capacity for representing ourselves impersonally. I then go on to make
the even stronger claim that the very notion of a representational
content that has any cognitive relevance is parasitic on our ability to
form an impersonal consciousness of self. Thus, even representations of
the world and the self that are independent of thought, representations
that Kant refers to as intuitions, have cognitive relevance for us only
insofar as we are able to take them as potential candidates for I thoughts.
This claim is the ultimate basis for the Kantian thesis that experience is
only intelligible to us to the extent that it is a potential content of
impersonal self-consciousness that is systematically linked to other po-
tential contents. It is also the basis for his famous thesis that there are
non-empirical conditions on all experience.
For Kant, non-empirical conditions on all experience are conditions
under which a self-conscious being is able to represent itself in any
arbitrary experience as the numerically identical point of view. This
representationof the self-consciousness as a numerically identical point of
view through di¬erent experiences connects di¬erent experiences to-
gether in a single possible representation. This representation of the self is
the same regardless of the di¬erent standpoints within experience that the
self-conscious individual might be occupying. In this way, the conditions
governing the representation of the numerical identity of the self pro-
vide one with constraints on the way that any objective experience must
be. And, insofar as these constraints also operate on one™s representation
of one™s personal identity as constituted by a certain sequence of points
of view within experience, they also provide the basis for an account of
 Kant and the demands of self-consciousness
©°  ® ¬  ®¤ °   ®¬  ¬ ¦ - ®   ©  µ  ®   
Personal self-consciousness involves an awareness of the distinction
between me and my representations and other persons and their repre-
sentations. In order for me to have some understanding of the distinc-
tion between me and my representations, and other persons and their
representations, I must have some way of comparing and contrasting
my identity as a person with a certain set of representations with that of
other possible persons with their own distinctive sets of representations.µ
In order to be able to compare and contrast my representations with
those of other persons, I must be able to abstract from the particular
identity, the particular set of beliefs and desires, that distinguishes me
from other persons. For I must be able to represent what it would be like
for me had I had a di¬erent set of representations than the ones that I
actually ascribe to myself:
It is obvious that: if one wants to represent a thinking being, one must put
oneself in its place, and place ones own subject under the object that one wants
to consider (which is not the case in any other kind of investigation), and that we
can only require an absolute unity of a subject for a thought because one could
not otherwise say: I think (the manifold in a representation). ( µ)*
The fact that I am able to represent the point of view of another rational
being does not mean that I am no longer the particular individual that I
am. But it does mean that I represent myself and other persons in an
impersonal manner. For, in representing what it might have been like
for things to appear to me in the way that they appear to the other being
to which I wish to attribute rationality, I represent myself as an arbitrary
self-consciousness, that is, just one person among many possible other
persons. But at the same time I am also able to represent myself as the
particular individual who I happen to be. For it is only in this way that I
can compare the representations that I might have had from the point of
view of another rational being with the representations that I have from
my own actual point of view.
If I come to have doubts about the states that I am ascribing to myself,
or if someone else challenges me concerning my past, I will feel the need
to consider the possibility that I might be mistaken in what states I think
* References to Kant™s Critique of Pure Reason (henceforth Critique) will be to the pagination of the ¬rst
and second editions of the Critique indicated by the letters A and B respectively. I follow the text
edited by Raymund Schmidt (Hamburg: Felix Meiner Verlag, ±°) except where otherwise noted.
All other citations of Kant™s work are based on the volume and page numbers of the critical edition
published by the Prussian Academy of Sciences and later by the German Academy of Sciences
(henceforth Ak.) (Berlin: de Gruyter: ±°“). Translations are mine throughout.
belong to my own history and even in who I am. I can only do so to the
extent that I am able to abstract from my actual personal identity, and
evaluate the reasons for ascribing certain states to myself in a manner
that would have weight for other persons as well. Thus, in order for each
of us to understand what it is to be a person with beliefs, emotions, and
desires, we must have an understanding of what it might have been like
to have a di¬erent set of beliefs, emotions, and desires. The possibility of
the point of view that we must take in order to go through these
alternative sets of beliefs, emotions, and desires gives self-consciousness
its transcendental dimension, that is, it makes self-consciousness a con-
dition under which we can recognize an object that is distinct from our
individual momentary representations of the world.
We can refer to the self that functions as a variable in self-conscious-
ness as the transcendental self:
We presuppose nothing other than the simple and in itself completely empty of
content representation: I; of which one cannot even say that it is a concept, but
rather a mere consciousness, that accompanies all concepts. Through this I, or
he, or it (the thing) that thinks nothing other than a transcendental subject = x is
represented. This transcendental subject is known only through the thoughts
that are its predicates. ( µ“/ °)
It might seem that the idea of a transcendental self commits one to a
featureless bearer of experience. But the dummy sortal x that stands in for
di¬erent individual constants would be misunderstood if taken to mean
that when we represent ourselves by means of I thoughts we are then
mere bare particulars, or egos bare of any properties that one could
come to know through experience. The notion of a transpersonal and
standpoint-neutral bearer of experience would be incoherent. In order
to be able to represent something, it would have to have some kind of
standpoint from which it represents things or at least some determinate
set of capacities with which it represents, but, in order to be a transper-
sonal and standpoint-neutral subject, it would have to have no proper-
ties in particular.
Fortunately, Kant does not think of the subject of transcendental
self-consciousness as a particular that has no particular properties,
although he thinks that this is a view to which Descartes was attracted in
trying to infer substantial properties of thinking beings in general from
the conditions under which we ascribe thoughts. For Kant, transcen-
dental self-consciousness is a representation of oneself that abstracts
from what distinguishes one from other persons, not a representation of
a bare particular:
 Kant and the demands of self-consciousness
It means a something in general (transcendental subject) the representation of
which must indeed be simple, precisely for this reason, since nothing is deter-
mined with respect to it, for certainly nothing simpler can be represented than
the concept of a mere something. The simplicity of the representation of a
subject is not therefore a cognition of the simplicity of the subject itself, for one
has completely abstracted from its properties, when it is merely designated by
the completely empty of content expression: I think (which I can apply to any
thinking subject). ( µ)

While I represent myself in a simple way when I represent myself by the
expression ˜˜I™™ or by means of the expression ˜˜I think,™™ and even
represent other thinkers simply when I represent them as individuals
that can potentially say of themselves ˜˜I think,™™ it would be a mistake to
infer from this that the ego that is the bearer of such I thoughts must
itself be a simple individual or bare particular.

© °   ® ¬  ¬¦-   ®   ©  µ  ®     ® ¤  µ ¤ §  ® 
The kind of self-consciousness expressed by the statement ˜˜I think p,™™
where p is any proposition, is, for Kant, the basis for all use of concepts,
judgments, and inferences. In using concepts, and making judgments
and inferences, we commit ourselves to a representation of what we are
representing by means of our concepts, judgments, or inferences that is
not just true for our own individual point of view, but is also true for any
arbitrary point of view. Kant refers to this notion of a representation
that is a representation for any arbitrary point of view as a representa-
tion that belongs to ˜˜a consciousness in general™™ (Bewußtsein uberhaupt), as
opposed to a representation that belongs to one consciousness alone.
Now Kant does not wish to argue that there are representations that
do not belong to the individual consciousness of distinct individuals. His
claim is rather that we understand what we are representing when we
are able to represent the content of representations that belong to our
individual consciousness in a way that, in principle, is also accessible to
other representers. The capacity to represent individual representations
in this manner that is accessible to other representers is just what Kant
regards as the capacity to use concepts. The capacity to use concepts is,
in turn, exhibited in the ability to make judgments that have determin-
able truth value, and to draw inferences on the basis of those judgments
that we can determine to be correct or incorrect.
In judgment, we may entertain the possibility that something is the
case, but we also commit ourselves to the assumption that what we judge
is or is not the case. This commitment expresses itself in a willingness to
o¬er reasons for our belief that something is or is not thus and such. In
taking on the obligation to o¬er reasons for what we judge to be the case,
we acknowledge that judgment is governed by normative principles.
These normative principles are based on the commitment to truth that
one takes on when one makes a judgment. Normative principles provide
procedures for distinguishing judgment that succeeds in articulating
truth from judgment that is false. These procedures may be articulated
in the form of rules governing the behavior of individuals. The norms
governing representation express themselves in terms of rules concern-
ing when to token a certain representation if we are to succeed in
articulating some truth. Our competence in judgment is then measured
against our ability to express truths by means of the judgments that we
Judgment actually presupposes both the kind of personal self-con-
sciousness that Kant refers to as empirical apperception and the imper-
sonal self-consciousness that he refers to as transcendental appercep-
tion. Judgment presupposes personal self-consciousness insofar as
judgment involves an implicit or explicit commitment on the part of the
person who forms the judgment that things are thus and such for him,
her, or it. At the same time, judgment also presupposes an impersonal
self-consciousness, for when one makes a judgment one makes an
assertion to the e¬ect that things are thus and such not only for one as
the particular individual that one is, but that, in principle, things should
be taken as thus and such by anyone.
At least some implicit consciousness of self is built into the normative
commitment that a judger takes on for her-, him-, or itself. To judge is to
place oneself in the space of reasons and thus to take on a commitment
to o¬er reasons for what one judges to be the case. But this means that,
in making a judgment, the judger implicitly takes her-, him-, or itself to
be not just conforming to rules but also tacitly or overtly obeying rules.
Kant links the capacity for obeying rules that we display in our ability to
use concepts to pick out and characterize objects not only with our
capacity for judgment, but also with our capacity for self-consciousness.
To have an idea that an individual is obeying rather than merely
conforming to norms of which s/he has no implicit or explicit under-
standing, we must regard her or his point of view as one that we might
be able to occupy in obeying the rules that we do. This is just to attribute
the capacity for self-consciousness to those creatures.
Bona ¬de norms must be principles that the individual can come to
 Kant and the demands of self-consciousness
understand as the basis for his or her behavior, and they must be
principles that the individual can come to see him- or herself as having
chosen to be bound by in his or her behavior. Such capacity for choice is
what Kant refers to as ˜˜spontaneity.™™ He regards it as a distinctive
feature of rational and hence self-conscious beings. Such creatures are
rational because they can assume responsibility for their own represen-
tations. It is this capacity to take responsibility that is the basis for their
possession of full-¬‚edged beliefs. To have full-¬‚edged beliefs, one must
be able to take something to be true. And, in order to be able to take
something to be true, one must be able to form one™s belief in accord-
ance with norms that licence one to take as true what one takes as true.
In forming a judgment, the individual is not merely stating a fact
about the way that individual interprets matters, the individual is also
making a claim that others ought to interpret things in the same way.
The individual is thus committing him-, her-, or itself to the possibility of
providing reasons for why he, she, or it has judged in that way rather
than in another way. These reasons operate as norms governing the
judgments in question. Norms are principles governing the responses of
individuals that apply to individuals in di¬erent situations.
Now it has often been claimed that normativity could stop at the level
of what a certain group or community takes to be true. While a view of
normativity that stops at the group allows for a shared communal point
of view relative to which individuals could be said to be right or wrong, it
fails to address the implicit claim of the group or community to articu-
late standards that hold for them not because they are the ones that they
do use but because those standards are the correct ones to adopt. A
con¬‚ict of belief or values between di¬erent communities is only intelli-
gible if the respective communities take themselves to be committed to
something that is not merely true or of value for them. Even if these
di¬erent communities see no way of establishing the validity of their
own point of view to the satisfaction of the other point of view, they still
must recognize the possibility of some encompassing perspective from
which their own view, in principle, could be justi¬ed. Thus, the norma-
tive commitment to truth requires the possibility of an impersonal point
of view, even if the point of view in question is not one that is ever
actually held by any person or group of persons.
Generalizing the point, we may say that, in order for one to be able to
recognize norms as norms governing one™s behavior, one must be able
to recognize principles that transcend a particular point of view. These
principles that transcend a particular point of view depend on one™s

ability to recognize not only one™s own point of view, but also the
possibility of other points of view to which those norms apply. For this,
one must have some understanding of what it would be like to be an
individual with such a distinct point of view governed by norms. But, in
order for one to be able to represent the possibility of another point of
view that is subject to the same principles to which one™s own point of
view is subject, one must be able to abstract from what is distinctive
about one™s own point of view. One must be able to place oneself in
thought or imagination in the position of another and re¬‚ect on what
things would be like from that alternative standpoint.
The self-consciousness expressed by the proposition ˜˜I think™™ pro-
vides each of us with an impersonal or, rather, transpersonal perspective
from which we are able to consider ourselves and others. The transper-
sonal perspective is just the way that we represent our own activities as
particular individuals to the extent that those activities are constrained
by norms that apply to absolutely all of us. These norms place us in the
space of reasons. This is why Kant insists that our only grip on the
notion of a rational being is through our ability to place ourselves in the
position of another creature. We are able to do this through the abstract
representation of self that we have in the self-consciousness expressed by
the proposition ˜˜I think.™™

 µ  ¬ © ®   ¦     § µ   ® 
My task in this book is ¬rst to show how Kant understands the notion of
transcendental self-consciousness. In the process, I distinguish his
understanding of this notion from the understanding of it provided by
other commentators. Then I develop the implications for an under-
standing of the general structure of experience that are inherent in the
notion of transcendental self-consciousness. I focus on the role that
transcendental self-consciousness has in connecting di¬erent spatial and
temporal episodes together in a single experience. This experience is
distinctive in that it is not the private experience of an individual, but, in
principle, is accessible to absolutely all of us. To clarify Kant™s concep-
tion of transcendental self-consciousness, I begin with a discussion of the
texts in the Critique of Pure Reason in which Kant ¬rst articulates the
notion of self-consciousness.
Kant introduces his distinction between empirical and non-empirical
self-consciousness in the ¬rst edition of the Transcendental Deduction
as a way of arguing for the claim that we have non-empirical concepts
±° Kant and the demands of self-consciousness
that may legitimately be applied to experience. In the A-Deduction,
Kant tries to establish that all contents of experience depend for their
very existence on the possibility of connecting them together in a
representation of self that is neutral with respect to the di¬erent contents
of experience. He argues that this is only possible to the extent to which
such contents of experience are subject to rules that connect those
representations together independently of experience. He refers to these
rules governing the possibility of an impersonal representation of self as
the categories of the pure understanding. The Transcendental Deduc-
tion is concerned with proving that such rules are bona ¬de rules in that
they must actually apply to all experience. In proving that there are
necessary and universally applicable rules governing experience, the
Deduction also provides a defense of objectivity. For such rules allow us
to form judgments about the objects of experience that must be true not
just for me or you, but for anyone.
In the next chapter, I argue that the notion of transcendental apper-
ception that is introduced in the A-Deduction is not to be understood as
a representation of personal identity. Instead, it is to be understood as a
condition under which it is possible for us to form concepts of objects. As
such, it is a representation of self that is the same for all of us. I criticize
contemporary interpretations of transcendental self-consciousness as a
kind of a priori certainty of personal identity, and argue that Kant was
not concerned with providing a direct response to Hume™s worries
about personal identity. Instead, Kant introduces his impersonal con-
sciousness of self as a condition for the formation of concepts of experi-
ence. I argue that the success of this argument depends on conceiving of
concept use and representation in general as representing the world in a
way that is the same for all individuals and that is also inherently
We represent items against a background of other representations
that give those representations their distinctive content. If representa-
tions are to belong together in an impersonal self-consciousness, they
must be connectable according to rules that allow us to represent
ourselves as having the same point of view irrespective of the di¬erences
in representational content that distinguish those representations from
each other. These rules have a cognitive content that is the same for all
of us under all circumstances because that cognitive content is deter-
mined by the inherently systematic and standpoint-neutral notion of
functional role in judgment and inference.
A number of contemporary interpreters have understood Kant to be
a functionalist about the self and the mind. I argue that Kant can only
be regarded as a functionalist in a very circumspect sense; he is con-
cerned with cognitive content as constituted by the functional role of
such content in judgment and inference. Thus, unlike most contempor-
ary functionalists, and contra most functionalist interpretations of Kant, I
argue that Kant only regards the mind as a functional system with
respect to the contribution of the active, spontaneous, aspect of the
mind, rather than with its passive dependence on causal relations
between representational contents.
In chapter three, I argue that Kant™s conception of the point of view
from which content is to be ascribed is based on his rejection of Hume™s
fundamental assumption that experience consists only of similarity
relations between numerically distinct perceivings. Kant argues that the
possibility of being conscious of one™s self-identity as a self-conscious
being is the basis for any conceptual recognition. He also plausibly
argues that conceptual recognition of an object must be possible if any
signi¬cant similarity relations are to be discerned. Without self-con-
sciousness one would not be able to distinguish a successful from an
unsuccessful recognition of an item by means of a concept, for one
would have no conception of the possibility that the item might present
itself to oneself in a way that is other than it is. And, without the
possibility of distinguishing unsuccessful from successful recognition,
there would be no basis for claiming that one had picked out relevant
similarities in experience either.
The associationist conception of experience developed by British
empiricism depends on the idea that we can have a brute recognition of
similarities without any underlying capacity for representing our ident-
ity as thinkers. I argue that Kant was right that this idea of brute
recognition will not work. The postulation of a brute capacity for
recognition fails to do justice to the normative character of recognition,
that is, that recognition can be successful or unsuccessful. Our associ-
ations cannot be completely random if they are to account for our
awareness of any regularities in experience.
I note that there are ¬rst-order rules that allow us to compare and
contrast various perceptual representations and represent them in a
standpoint-neutral way. These rules are what Kant calls empirical
concepts. There are also, however, second-order non-empirical con-
cepts that make it possible for us to form empirical concepts. These
second-order concepts dictate that nature must have the kind of uni-
formity that allows one to connect distinct representations together in
± Kant and the demands of self-consciousness
one possible self-conscious experience. They are what Kant refers to as
the categories. The categories are su¬cient to establish a general uni-
formity in nature. But they do not tell us what particular form such
uniformity must take. They do not tell us which particular laws nature
must obey.
This is why our ability to apply second-order concepts or categories to
experience is governed by still higher-order concepts, which Kant refers
to as ideas of reason. Such ideas of reason project a certain kind of
systematic unity onto the whole of nature and thus allow us to identify
the particular forms of regularity required for the formation of particu-
lar empirical concepts. We apply concepts to experience in ways that
always involve some implicit commitments to how other concepts are to
be understood. It is only through such systematic representational
commitments that we are able to distinguish representations that are
true of their objects from those that are not. For our only grip on objects
that are independent of us is through our capacity systematically to
apply the concepts that we have to experience. We have this capacity
systematically to articulate and apply concepts because we are able to
connect di¬erent concepts together in an impersonal representation of
their di¬erent contents that expresses what they ought to represent for
In chapter four, I take up the relation of thought and judgment to the
self-consciousness expressed in the proposition ˜˜I think.™™ Here, I focus
on the revised argument of the B-Deduction.The B-Deduction makes
the connection between being a potential candidate for impersonal
self-consciousness and being a potential candidate for judgment explicit
in a way that is lacking in the A-Deduction. First, I note the importance
of the proposition ˜˜I think™™ for cognitively relevant content. I note that
contents of representation are cognitively relevant to us inasmuch as
they can be thought by us. This means that contents of representation
are cognitively signi¬cant for us insofar as they are potential candidates
for judgment. I then develop Kant™s argument that anything that can be
thought by us has a relation to a possible self-consciousness ˜˜I think™™ in
virtue of the enabling role of such self-consciousness in the formation of
concepts and judgments.
Representations have relations to each other that are based on the
identity and di¬erences between the objects that they represent. The
most crucial of these relations are ones that preserve the truth of a
representation. Here, the truth of a representation consists in a repre-
sentation representing its intended object as that object is independently
of that representation. Truth is particularly what is at issue when we
make a judgment or claim. And truth is preserved between the contents
of representations by means of logical relations. These logical relations
constitute the most general conditions under which we can ascribe
content to representations. These most general conditions for content
ascription are the most abstract conceptual conditions governing the
possibility of self-consciousness.
I argue that the key to an understanding of the intellectual pre-
conditions on representation is the constitutive role that both personal
(empirical) and impersonal (transcendental) consciousness of self play in
our capacity to form concepts and articulate them in judgments. Any-
thing that is to be a concept must be such that it is capable of articulating
some content in a way that is in principle accessible to any one of us and,
indeed, all of us. This capacity to represent things in a person-neutral
way needs to be displayed in judgments that have a truth value that
purports to be independent of the way a particular individual happens
to respond to a particular situation. In judgments, we are able to use
concepts to make objective claims that purport to be true not only for
me or you, but for anyone.
Kant maintains that representations must be potential candidates for
inclusion in a consciousness of oneself that potentially includes all
possible representations; This universal self-consciousness is a possible
although never actual co-consciousness of all of one™s representations.
One never actually surveys all of one™s representations, much less all
possible representations; instead one is able to represent their distinctive
contents by connecting them according to rules that have an implicit
reference back to oneself as subject of thought. This implicit self-
reference is needed for rules constituting the cognitive signi¬cance of
various contents, because representations have cognitive signi¬cance
only to the extent that they are potential candidates for comparison and
contrast by some subject. To be compared and contrasted by a subject
they must present themselves to that subject, and, as such, they must be
something for that subject. The demand that all representations be
potential candidates for self-consciousness is the basis for a claim that all
represented objects stand under the normative constraint of being
potential objects of judgment. As objects of judgment that purport to
have objective validity, represented objects may be regarded as objec-
tive. Even judgments concerning subjective states must have objective
import; this leads to the problem of how to ¬nd a place for knowledge of
subjective states.
± Kant and the demands of self-consciousness
In chapter ¬ve, I argue that Kant is forced to introduce a second step
in the proof to explain how knowledge of even subjective states is
possible. He ¬rst argues that our knowledge of objects is restricted to
spatio-temporal objects. Then, he argues that even our inner states that
are temporal depend on the existence of outer states that are spatial.
The dependence of inner experience on outer experience allows him to
argue that even our perceptions and other inner episodes are subject to
the same necessary conditions to which intersubjectively available ob-
jects must be subject. This is because even our perceptions provide us
with a way of representing the spatio-temporal world from a certain
point of view only because they can be integrated into an impersonal
and hence objective way of representing the spatio-temporal world for
any arbitrary perceiver. The key here is to understand the manner in
which not only empirical self-consciousness, but also representation in
general, depend on transcendental self-consciousness and thus allow for
judgments concerning even one™s subjective states.
The argument that self-consciousness is a source of substantive con-
straints on experience depends on something more than the very gen-
eral idea that we are capable of forming concepts and making judg-
ments. Kant™s argument for objectivity from the postulation of a non-
empirical self-consciousness depends essentially on the assumption that
we must represent the world temporally because this is constitutive of
our very conception of what is internal to our own point of view.
Non-empirical consciousness of self is introduced as an enabling condi-
tion of our necessary temporal representation of our experiences.
The idea that all experiences have a temporal structure must be
linked to more general conceptual constraints on experience. First it
must be seen that we are able to think of representations as being in time
because we can order those representations in such a way that we can
ascribe them to di¬erent individuals who have sets of experiences that
constitute di¬erent temporal series. These di¬erent temporal series can
only be compared and contrasted with each other to the extent that they
may be regarded as belonging to a single shared time. This single shared
time is the temporal form that di¬erent experiences have in virtue of
belonging to one possible impersonal self-consciousness.
The only way we can account for the regularities in what we perceive
is in terms of the assumption that what we are perceiving is connected to
what we would perceive from a di¬erent spatio-temporal point of view
according to laws. It is di¬cult, if not impossible, to identify any laws
connecting sense perception to various kinds of objects. The laws in
question must therefore be laws governing the objects that we perceive
independently of their being perceived. The problem here is that we
have knowledge of the objects perceived only through our perceptions.
Kant argues that this problem can be resolved once we realize that the
laws governing the objects perceived and indeed governing our associ-
ations of di¬erent perceptions are nothing but the uni¬ability of di¬er-
ent perceptions in an impersonal self-consciousness. This uni¬ability of
perceptions in an impersonal self-consciousness is just the idea that
di¬erent perceptions are connected in an individual consciousness in the
same way that they ought to be connected in any consciousness that
perceives or represents things as they are independently of that con-
sciousness.The regularities in experience that present themselves to all
of us as self-conscious beings re¬‚ect our ability to combine representa-
tions together in consciousness in a manner that is not unique to each
individual. It is in virtue of such impersonal consciousness of self that we
are able to form empirical concepts of the objects that we perceive and
are then able to apply those concepts to what we perceive.
In chapter six, I discuss the theory of time-determination developed
in the Analogies of Experience. It works out the implications of the idea
adumbrated in the Deduction that the unity of space and time (as forms
according to which we distinguish the outer from the inner) is a function
of the systematic relations that the di¬erent spaces and times represen-
ted by di¬erent possible individuals have to a possible self-conscious-
ness. Kant™s general idea that spatio-temporal representations must
make a di¬erential contribution to consciousness if they are to belong to
the experiences of a self-conscious being is the basis for the general
assumption of the Analogies that times and spaces must be empirically
distinguishable. In the First Analogy, Kant defends the need to postulate
sempiternal substances as the basis for recognizing changes in objects of
experience. These substances underwrite our ability to ascribe a deter-
minate position in time and space to representations and objects repre-
sented by us. For we have knowledge of positions in time and space only
through di¬erences that can be made out in what we experience. These
di¬erences manifest themselves temporally in the di¬erences between
events. Kant argues that these di¬erences between events are to be
interpreted as changes in the states of things. He can claim that all
changes must be recognizable in experience on the basis of his robust
theory of transcendental idealism. For this robust theory of transcen-
dental idealism does not allow for radically mind-independent and
hence recognition-independent events. Even without this strong version
± Kant and the demands of self-consciousness
of transcendental idealism, a case can be made for the need to presup-
pose persistent substances if changes are to be recognizable. However, it
cannot be demonstrated that events must be recognizable except insofar
as they are to be objects of our experience.
Kant™s defense of the general causal principle is based on the idea that
the temporal order of episodes in any change must be empirically
determinable. It thus builds on the necessity of the recognizability of
change argued for in the First Analogy on the basis of the principle that
empirical representations must make a determinable di¬erence to ex-
perience if they are to be potential candidates for self-consciousness.
While Kant rejects the causal theory of time when it is understood to
reduce the meaning of temporal terms to causal relations, he argues that
causation allows one to determine which of two events occurred earlier
and which occurred later.
In chapter seven, I discuss the relation of the general causal principle
and the general principle that there must be substances and interactions
in experience, to our capacity to formulate speci¬c laws governing
causation, interaction, and individual things. The only way we can
know that a speci¬c change from event-type A to event-type B has
occurred and thus that A must precede B is if this change follows in a
lawlike fashion upon some other event type of which we have knowl-
edge. Such lawlike succession is just what we mean by causal connec-
tion. Interactions between substances are the basis for our knowledge of
simultaneity relations between those substances. By being able to deter-
mine the temporal order of what is represented by us, we are able to give
empirical content to distinctions between di¬erent spatial and temporal
points of view. At the same time, we are able to connect anything that is
represented by us together with anything else that is represented by us in
a single consciousness of the temporal unity and the di¬erences of
empirical points of view. Kant seems to think that causation and
interaction can only assign determinate temporal positions to objects
and events if they are capable of providing su¬cient conditions for
change. However, he allows for indeterministic causal laws at the level
of human action, and, in the light of current fundamental physical
theory, it seems more plausible to weaken this assumption so that
probabilistic laws governing causal connections and interactions be-
come possible at the level of fundamental natural processes. In the
concluding sections of the chapter, I argue that Kant™s account of causal
laws is compatible with free action. The application of causal laws is
governed by causal conditions that we assume to comprise a complete
set for the regulative purposes of inquiry. However, the important point
to see is that we never are in fact capable, even in principle, of ident-
ifying a complete set of such causal conditions. This always leaves space
for an alternative account of human action under action descriptions
that are independent of actual causal conditions.
After discussing the general relation of substance, cause, and interac-
tion to particular kinds of substances, causes, and interactions in chapter
seven, I turn in chapter eight to the temptation to think of the self as a
thinking thing that is a substance endowed with personal identity over
all time. This temptation or ˜˜transcendental illusion,™™ as Kant calls it, is
rooted in the nature of our access to the self from the ¬rst-person point
of view. Because I thoughts are self-verifying thoughts, and because we
have access to other rational beings by thinking of them as if we were in
their place as I thinkers, we become tempted to think that the ¬rst-
person point of view of self-consciousness is capable of communicating
substantive truths about the nature of thinking beings in general. In
chapter nine, I look at how this essentially ¬rst-person access to rational
beings encourages us to think of ourselves as substances that are inde-
pendent of material objects and knowable in a more certain way than
things that exist outside of us.
In chapter ten, I discuss Kant™s refutation of idealism which is a
revised version of his critique of the kind of epistemic dualism that takes
our knowledge of our inner states to be more certain than our knowl-
edge of outer states. Kant maintains that at least some of the objects that
we directly experience must be outside of us in space. He attempts to
establish this claim by means of an argument showing that determinate
consciousness of one™s own inner experience is only possible if there are
actual outer objects. The argument thus establishes a necessary link
between what can be regarded as internal to the point of view of a
particular self-conscious being and what can be regarded as external to
the point of view of a particular self-conscious being.
The argument against the kind of skepticism about the existence of
the external world that Descartes articulates in his First Meditation is
based on the general thesis that one cannot ascribe determinate beliefs
to oneself without being able to order those beliefs in a determinate
temporal order. It is then argued that one cannot ascribe a determinate
temporal order to one™s beliefs without some direct consciousness of
something that is not inherently successive. One™s occurrent beliefs and
desires are inherently successive. They pick out di¬erent nows of aware-
ness due to their character as di¬erent occurrent states of awareness. I
± Kant and the demands of self-consciousness
maintain that Kant needs to be taken at his word that any determinate
consciousness of oneself requires an immediate relation to something
outside of that self-consciousness. I contend that the argument against
˜˜psychological™™ idealism has force against the Cartesian skeptic who
already accepts the possibility of self-knowledge.
The refutation is not complete until it addresses the manner in which
our beliefs depend on not only objects that are outside of us in the
empirical, but also things in themselves that are outside of us in the sense
of being completely independent of our minds. This ultimately leads
Kant to raise the issue of transcendental idealism in coming to terms
with the problem of how to refute idealism.
I take up transcendental idealism in chapter eleven. Transcendental
idealism is the thesis that the only objects of which we can have
substantive representations are objects as they must appear to us accord-
ing to our a priori forms of sensibility. Sensible pre-conditions on
experience restrict our experience to objects as they must appear to us,
rather than allowing us access to things as they are independently of the
way we must represent them as internal to, or external to, our point of
view. I argue that Kant vacillates between a modest version of transcen-
dental idealism according to which we cannot resolve the question of
what the ultimate nature is of objects that are independent of the pre-
conditions that we bring to experience, and a more ambitious claim that
objects as they are independently of our experience cannot be spatial or
temporal at all. Only the former idealism seems to me to be defensible.
In relating Kant™s argument for transcendental idealism to his argument
against empirical or psychological idealism, I discuss some of Kant™s
personal notes (his so-called ˜˜re¬‚ections™™) which I try to handle with
care, since they cannot claim the same authority as the material that he
chose to publish. I conclude with a discussion of the general account of
experience implied by my reconstruction.
° 

Introducing apperception

Kant introduces the notion of apperception as well as the notion of
self-consciousness in the Transcendental Deduction of the Categories.
This is the text of the Critique of Pure Reason that is widely regarded as the
most central one of the whole Critique. It is the section of the Critique that
Kant said had cost him ˜˜the most trouble,™™ presumably because it is
˜˜laid out at a rather deep level™™ ( ©).
In this chapter, I propose to develop Kant™s account of apperception
and the general way in which Kant connects apperception to represen-
tational content in the ¬rst (A) edition of the Deduction. First, I argue
that the A-Deduction interprets the notion of apperception as self-
consciousness. I then argue that the numerical identity that Kant
ascribes to transcendental self-consciousness in the A-Deduction is not
to be understood as committing him to any speci¬c claims about my
individual personal identity. It is rather to be understood as an enabling
condition of conceptual recognition of objects. I discuss Kant™s argu-
ment that all representational content must have at least an indirect
relation to a possible self-consciousness in order to be a determinate
representation at all. I argue that this is best understood as the idea that
each representation has a distinctive functional role in judgment and
inference that is based on its relation to a possible self-consciousness.

° °   ° ©  ® © ®    -¤  ¤ µ  ©  ®
Kant introduces empirical apperception in the following way:
The consciousness of oneself according to the determination of our state in
inner perception is merely empirical, always mutable, there can be no standing
or persistent self in the ¬‚ux of these inner appearances, and it is customarily
called inner sense, or empirical apperception. ( ±°·)
Here, Kant identi¬es empirical self-consciousness with inner sense and
empirical apperception. In interpreting empirical self-consciousness in
° Kant and the demands of self-consciousness
terms of the notions of inner sense and apperception, he appeals to
accepted terminology in the Leibnizian tradition. Philosophers in that
tradition identi¬ed inner sense with empirical apperception.± They took
inner sense to be an experience of inner states, while they took empirical
apperception to be a consciousness of inner states. Since they took inner
states to be states of consciousness, they regarded empirical appercep-
tion as a form of self-consciousness, that is, they took empirical apper-
ception to be a consciousness of perceptual consciousness. For Kant
˜˜inner sense [is that] by means of which the mind intuits itself and its
state™™( / ·), where ˜˜time is nothing but the form of inner sense,
that is, of the intuiting of ourselves and our inner state™™ ( / ).
The implication of these passages is that empirical self-consciousness
involves some kind of intuition. Since Kant de¬nes an intuition as a
representation ˜˜that relates immediately to an object and is singular™™ (
°/ ··), the implication is that empirical self-consciousness is an
immediate consciousness of oneself as an individual. This immediate
representation is a representation of oneself at a certain time, but it is not
a representation of oneself over time.
Empirical apperception represents the self in terms of the individual
states that replace themselves in the succession of di¬erent states of
consciousness in time. When we are conscious of our inner empirical
states, we are not conscious of anything that is identical over time. But it
is important to note that Kant does not deny that we have an intuition of
self through inner sense and that we therefore have some consciousness
of self even in empirical self-consciousness. The important point is that
we do not directly experience anything as something connecting our
various experiences together in time.
The ephemerity of mental episodes encourages Kant to argue that in
order to have a representation of self as the identical subject of di¬erent
experiences we must be able to represent something that is not given in
any experience. Thus, it might seem that we could avoid the conclusion
that we need something non-empirical to serve as an identical self by
appealing to an object of outer experience as the experientially access-
ible bearer of experiences. One might argue that in proprioception of
our bodies, that is, in our immediate experience of our bodies, we have a
direct representation of an embodied self. But even proprioception
provides at best synchronous consciousness of self; it fails to provide us
with a representation of our own identity over di¬erent times and
spaces. And most signi¬cantly, proprioception does not provide us with
the kind of necessary representation of numerical identity in which Kant
Introducing apperception
is interested. For his real concern is not with how an identical self can be
represented, but rather with how something could be represented as
necessarily identical in di¬erent experiences:

That which should necessarily be represented as numerically identical cannot be
thought as such through empirical data. It must be a condition that precedes all
experience and even makes it possible that validates such a transcendental
condition. ( ±°·)

Several questions arise at this point: (±) What is a transcendental condi-
tion? () What is the transcendental condition in question? And () for
what is the transcendental condition a condition? Perhaps Kant™s best
answer to the question of what a transcendental condition is, comes in a
discussion of apperception in the section criticizing the false inferences
or paralogisms of rational psychology:

For this inner perception is nothing more than the mere apperception: I think;
which even makes all transcendental concepts possible, in which it is said: I
think substance, cause, etc. For inner perception in general and its possibility,
or perception in general, and its relationship to other perception without a
particular di¬erence between perceptions and determination being given em-
pirically, cannot be regarded as empirical, but must be regarded as cognition of
the empirical, and belongs to the investigation of the possibility of any cogni-
tion, which is indeed transcendental. ( / °±)

In other words, a transcendental condition is a condition under which
cognition in general, and empirical cognition in particular, is possible.
Kant regards transcendental apperception as such a non-empirical
condition on what can be known empirically. Indeed, Kant identi¬es
the transcendental condition in which he is interested in the A-Deduc-
tion as transcendental apperception. In the A-Deduction, he maintains
that transcendental apperception makes it possible to explain the exist-
ence of a necessary connection between representations which he ar-
gues is involved in any empirical cognition. For this necessary connec-
tion is supposed to be nothing but the concept of an object that
corresponds to our representations. This provides a general answer to
the third question: of what is transcendental apperception a transcen-
dental condition? Transcendental apperception is a condition under
which it is possible to have a concept or cognition of an object. What the
relationship between concepts and cognitions is supposed to be, and
why we should understand the concept of an object in the way that Kant
proposes, will have to be determined later. For now it is su¬cient to note
 Kant and the demands of self-consciousness
that transcendental apperception is supposed to account for our capac-
ity to form concepts of objects:

This necessity has a transcendental condition as its ground. Therefore a
transcendental ground must be found for the unity of consciousness in the
synthesis of the manifold of all our intuitions, hence also of all concepts of
objects in general, consequently of all objects of experience, without which it
would be impossible to think any object; for this [object] is nothing more than
the something the concept of which expresses such a necessity of synthesis. This
original and transcendental condition is no other than transcendental appercep-
tion. ( ±°·)

When Kant introduces transcendental apperception as a necessary
representation of numerical identity, he does not explicitly say that the
numerical identity that he is concerned with is that of the self. This has
led Andrew Brook to argue that the empirical and transcendental
apperception to which Kant refers at  ±°· is not a consciousness of self
at all, but merely awareness of something. Now Kant explicitly claims
that empirical apperception is a kind of self-consciousness. The context
also suggests that Kant thinks that the standing self that he misses in
empirical apperception must be supplied by a transcendental represen-
tation. And he later clearly states that numerical identity is certain a
priori with respect to all possible self-consciousness, since nothing can
enter cognition except via this original apperception ( ±±). Moreover,
Kant also talks of an ˜˜original and necessary consciousness of the
identity of oneself™™ ( ±°), as what makes it possible for us to determine
an object for our experiences. If this is not enough evidence, Kant also
speaks of ˜˜the proposition that expresses self-consciousness: I think™™ (
“), after already talking of ˜˜the mere apperception: I think™™ (
/ °). So it is quite implausible to argue, as Brook does, that
apperception is not self-consciousness and that, therefore, self-con-
sciousness is not crucial to Kant™s argument.
Kant™s talk of a representation of numerical identity with respect to
transcendental apperception encourages one to think of transcendental
self-consciousness as consciousness of personal identity in contrast with
consciousness of individual states involved in empirical apperception.
But, in fact, the notion of transcendental self-consciousness is imperso-
nal in a way that, in principle, is in transpersonal. The necessity of
representing oneself as numerically identical does not commit Kant to
the existence of a persistent bearer of my states of consciousness, but
rather to a way of representing ourselves, a point of view from which
Introducing apperception
what is represented by me and you at di¬erent times and places can be
uni¬ed. Kant maintains that the self is necessarily represented as numeri-
cally identical; he does not argue that it is necessarily numerically
identical over di¬erent states.
But how are we to understand the necessity of representing ourselves
as numerically identical? And in what sense can we talk of a certainty
that our self-consciousness is numerically identical? All of us use the
expression ˜˜I™™ to refer to ourselves. The role of the demonstrative
expression ˜˜I™™ in connecting together representations that individuals
have of themselves at di¬erent times and spaces in one unitary experi-
ence is what makes for the numerical identity of our representation of I.
The expression ˜˜I™™ articulates a self-consciousness that remains the
numerically same point of view regardless of one™s spatio-temporal
situation. Now I might be mistaken about who I am, and thus about my
personal identity, but not about the fact that I am now self-conscious
and that I can represent myself as the same subject in alternative
situations. Thus, even though Kant thinks of transcendental self-con-
sciousness as necessary to any consciousness of an object, he does not
regard transcendental self-consciousness itself as a personal self-con-
sciousness at all. The necessary representation of numerical identity in
transcendental self-consciousness is, rather, the necessary representa-
tion of a shared point of view from which we can make sense of an
objective space and time and, indeed, of the communicability of the
contents of concepts to di¬erent spatio-temporal points of view:

This pure, original, unchanging consciousness I will now call transcendental
apperception. That it deserves this name is already clear from the following: that
even the purest objective unity, namely of concepts a priori (space and time) is
only possible through the relation of intuitions to it. The numerical unity of this
apperception lies a priori as much at the basis a priori of all concepts as the
manifold of space and time does of all intuitions of sensibility. ( ±°·)

The purity of self-consciousness refers to its independence from the
content of any particular experience. The original character of self-
consciousness is based on the idea that any personal or empirical
self-consciousness will depend for its existence on the possibility of that
impersonal consciousness of self. The unchanging character of such
consciousness is based on the fact that it represents a point of view that
must be regarded as identical in any experience. This point of view is the
basis for our ability to interpret experience in terms of concepts of
 Kant and the demands of self-consciousness
Kant insists that the ˜˜numerical unity of this apperception is the
ground a priori of all concepts™™ ( ±°·). The capacity for consciousness
of self-identity makes concepts possible by providing the idea of a
representer and hence a representation of that representer that is
distinguishable from what is represented and yet represents the repre-
senter as an I that, in principle, can be regarded as the possessor of an
arbitrary spatio-temporal point of view. Our concepts of space and time
have ˜˜objective unity™™ insofar as they capture the way a self-conscious
being would represent the world in the same way from any arbitrary
standpoint in space and time and in any arbitrary psychological state
that the representer might happen to be in.
Concepts are just the way in which things are represented in universal
terms, that is, represented in a way that connects di¬erent experiences
together in the same way for di¬erent persons in di¬erent psychological
states and situations: ˜˜All cognition demands a concept but this [con-
cept] is always something universal according to its form, and something
that serves as a rule™™ ( ±°). Now ˜˜concepts are based on functions™™ (
/ ), where the functions in question are the functional relations
involved in judgments that subsume one representation under another
representation: ˜˜But I understand under function the unity of the act of
subsuming di¬erent representations under a common one™™ ( / ).
The representation under which another representation is to be sub-
sumed is one that represents a feature that the ¬rst representation has in
common with other representations.
The representation of the common or shared feature is what Kant
calls a concept. Concepts, then, have their distinctive cognitive content
in virtue of the distinctive functional role that they play in the forming of
judgments and the drawing of inferences from those judgments. But
what all concepts have in common is that they are representations that
play identical functional roles despite di¬erences in the inner states of
the di¬erent individuals who use those concepts in di¬erent situations.
Kant notes that the kind of unity of consciousness that is displayed
by our capacity for conceptual recognition would be impossible ˜˜if the
mind in cognition of the manifold could not become conscious of the
identity of function through which it synthetically connects that [mani-
fold] together™™ ( ±°). From the need to be able to represent an
identity of functional role in di¬erent contents of experiences in order
to be able to recognize items in experience, Kant signi¬cantly con-
cludes that the ˜˜original and necessary consciousness of the identity of
oneself is at the same time a consciousness of the necessary unity of
Introducing apperception
synthesis of all appearances according to concepts™™ ( ±°). In other
words, for Kant, the necessary representation of the numerical identity
of the self is built into our ability to represent things in di¬erent
situations in ways that have the same cognitive role in judgment and
inference for all of us.
Necessary consciousness of our self-identity is something more than
the ability to represent ourselves as having a point of view from which
things appear in the same way to each and all of us. It is the capacity at
the same time to represent our point of view as the same point of view as
we compare di¬erent items of experience with respect to their identity
and di¬erences. It is only in this way that we are able to form concepts
with di¬erent cognitive roles because they have di¬erent functional roles
in judgment and inference. In sum, Kant links the necessity that we
represent ourselves as numerically identical in di¬erent experiences to
the possibility of forming an impersonal point of view. By taking this
impersonal point of view on our experiences, we are then able to form
concepts that have the same distinctive functional role in di¬erent
experiences. The distinctive functional roles that di¬erent concepts
have, in turn re¬‚ect the di¬erent systematic contributions that di¬erent
concepts make to the understanding of what we experience.

© ® °   ©®§   ®µ   ©  ¬ © ¤® © ¦   ¬ ¦
The ˜˜impersonal™™ character of our representation of self as numerically
identical has been generally obscured by contemporary obsession with
viewing Kant™s conception of the identity of apperception as a direct
response to Hume™s critique of personal identity. Hume™s reservations
about our knowledge of personal identity are summed up in the follow-
ing passage:
It must be some one impression, that gives rise to every real idea. But self or
person is not any one impression, but that to which our several impressions and
ideas are suppos™d to have a reference. If any impression gives rise to the idea of
self, that impression must continue invariable the same, thro™ the whole course
of our lives; since self is suppos™d to exist after that matter. But there is no
impression constant and invariable. Pain and pleasure, grief and joy, passions
and sensations succeed each other, and never all exist at the same time. It
cannot, therefore, be from any of these impressions, or from any other, that the
idea of self is deriv™d; and consequently there is no such idea.
Hume™s worry is that the representations (impressions) that make up
our mental life are continually replacing each other, so that there does
 Kant and the demands of self-consciousness
not seem to be any representation to which we could appeal in order to
provide a conception of our personal identity. Kant never attempts to
show that there is a ˜˜real idea™™ of the self, nor does he attempt to ¬nd an
impression from which such a representation could be derived. He thus
does not attempt to meet Hume™s demands with respect to personal
identity. Kant takes it as a given that we have consciousness of ourselves
with respect to our inner states even though the self of this inner
experience is in a perpetual state of ¬‚ux. But he also agrees with Hume™s
worries about the introspective basis for consciousness of one™s identity
over time. Given Kant™s view that we have an intuition of self in inner
sense, I see no reason to saddle him with Hume™s view that the self is
never itself an object of perception or introspection. However, such a
claim would only provide further support for Kant™s assumption that the
self is accessible only through the formal structure that is inherent in the
activity of self-consciousness.
Interpreting Kant™s notion that we necessarily represent ourselves as
self-identical a priori as a direct response to Hume™s worries about
self-identity forces one to identify a priori consciousness of self-identity
with a priori knowledge of one™s identity as a discrete individual. This is
very di¬cult to reconcile with Kant™s claim that self-consciousness
˜˜according to the determinations of our state™™ is empirical and does not
have a persistent subject ( ±°·). It also ignores the fact that Kant
commits himself only to an a priori representation of numerical identity,
not a priori knowledge that we are numerically identical.
Not all of the di¬culties with Kant™s notion of a necessary representa-
tion of self as numerically identical can be traced back to regarding Kant
as a direct respondent to Hume™s worries about our knowledge of
personal identity. Much of this controversy has however been generated
by the more general conviction that Kant™s talk of the numerical identity
of the self must commit him to the claim that I can know that I am the
numerically same person through di¬erent experiences. It should not be
denied that Kant™s talk of numerical identity of the self encourages an
interpretation that identi¬es consciousness of numerical identity with
knowledge of personal identity. This is part of the reason that it has
seemed unclear to commentators whether Kant is concerned in his talk
of numerical identity of the self with the identity of a person or with a
person™s being conscious of identical thoughts.µ
Elsewhere in the A-Deduction, Kant interprets the numerical identity
that we necessarily represent, as a numerical identity of possible self-
consciousness. This numerical identity is characterized as a priori cer-
Introducing apperception
tain. Kant makes remarks that have seemed to interpreters to suggest
that we might be certain a priori of our numerical identity as persons as
a pre-condition for having representations at all:
All possible appearances belong as representations to the whole of possible
self-consciousness. But numerical identity is indivisible from it and a priori
certain because nothing can come into cognition except by means of this
original apperception. ( ±±)

In response to passages such as these in the A-Deduction in which
Kant ascribes a priori certainty to the numerical identity of self-con-
sciousness, Dieter Henrich has emphasized the fact that we have cri-
terialess consciousness of self-identity. This consciousness of self-identity
is supposed to be characterized by a certainty of the kind Descartes
discovered in I thoughts. Descartes argues famously that I thoughts are
self-verifying, to have those thoughts is already to have su¬cient war-
rant for regarding them as true. But he never argues that this self-
verifying character of I thoughts extends to claims that the self is
identical over time. However, Henrich™s idea that we have a Cartesian
certainty of the identity of the self over a sequence of states has an
antecedent in Strawson™s view that at the heart of the Cartesian illusion
that we can infer substantial facts about the self from I thoughts is the
fact that we have a criterialess consciousness of self in immediate or
recalled experience. There is just no question for me that the states that
I ascribe to my present and past consciousness actually belong to my
consciousness of self.
However, unlike Strawson, Henrich seems to take Cartesian certainty
in the direction of making a self-justifying claim about personal identity.
In this way, Henrich™s notion that we have Cartesian certainty of our
numerical identity through the transitions involved in understanding
the di¬erent aspects of objects seems to commit him to a priori knowl-
edge of the real persistence of a self.· For Henrich insists that we are
certain of our numerical identity through changes in states. And only
with respect to the real persistence of a self does it make sense to make
claims about changes in state.
Henrich argues that the self as subject of self-consciousness can only
be weakly or moderately, rather than strictly, identical with itself. For
self-consciousness involves acts of consciousness that are changes in state
and only the notion of a weak numerical identity of the self is supposed
to be consistent with changes in state. On the other hand, Henrich
thinks of such identity as identity through atemporal change. But
 Kant and the demands of self-consciousness
neither the idea that strict identity precludes change nor the idea that
atemporal changes are possible has any obvious support in the Kantian
text. And both ideas seem to be intrinsically quite implausible.
In response to criticism, in particular from Paul Guyer, that Henrich
simply assumes one™s self-identity over representational states as a syn-
thetic a priori premise governing the self-ascription of representation,
Henrich has distinguished his conception that one is certain of the
identity of one™s self through di¬erent states of self-consciousness from
empirical knowledge of personal identity as well as from the notion of
identity familiar from logic. Unfortunately, Henrich fails to give a
positive characterization of his conception of the identity implied in
self-consciousness. This would be less problematic if one could see what
alternative to the logical notion of identity there could be. Henrich is
probably tempted to argue that there is a non-logical notion of identity
by talk of a ˜˜loose™™ as opposed to a ˜˜strict™™ meaning of personal
identity. But, properly understood, the ˜˜loose™™ notion of identity in-
volved in such contexts is not really a distinctive notion of identity at all.
Philosophers who take the identity of a person to be ˜˜loose™™ regard
persons as constituted by a series of individual person-stages rather than
being the numerically same individual through the time of their lives.
But even if persons were mere series of person-stages, persons would
nevertheless be ˜˜strictly™™ identical through time, for they would be the
same series throughout their existence.
While Guyer criticizes Henrich for endorsing a kind of a priori
certainty of personal identity, he thinks that Henrich is right to attribute
to Kant the view that we have a kind of a priori knowledge of our
personal identity through di¬erent states. Indeed, Guyer thinks that
Kant assumes that we can be certain that all of our empirical states are
ones of which we can become conscious. This is because Guyer inter-
prets Kant™s principle that one can be conscious of one™s numerical
identity as subject of self-consciousness in respect to all possible repre-
sentations as a claim that one can have a priori certainty of all of one™s
representational states and hence of one™s personal identity as an em-
pirically knowable individual:

Kant has failed to establish that I must in fact know“a fortiori be certain“that I
have really had all of a putative series of representations through some period of
my continued existence in order to investigate their possible empirical signi¬-
cance. But unless Kant can exclude a priori the possibility that one of the results
of my investigation could be the very rejection of the supposition that I actually
Introducing apperception
had one or more of the representations the possible empirical connections of
which I am investigating, he cannot prove that certainty of my possession of any
particular representations really is presupposed by any empirical investigation
of them.
Guyer rightly disparages the idea that we could be certain that each
of the representations that we think we have had are in fact our own.
This would entail a priori certainty that all of the beliefs that we ascribe
to ourselves really are the beliefs that have belonged to our lives, and this
would entail a priori certainty of personal identity. But this interpreta-
tion of a priori certainty, as a priori certainty of one™s personal identity,
is based on his assimilation of Kant™s notion of transcendental self-
consciousness to consciousness of one™s self-identity as empirically
knowable, that is, to consciousness of one™s individual personal identity.
We can hardly rule out a priori that some of the claims to empirical
knowledge of our personal identity that we make might be false.
One striking feature of Guyer™s interpretation is that it is based on a
notion of a priori certainty of self-identity that Kant introduces to
account for our capacity to recognize objects that are distinct from our
momentary present states of consciousness. While Guyer rejects such a
priori certainty, he has made the synthesis of apprehension construed as
the interpretation of momentary intuitions of multiplicities the key to his
interpretation of the Transcendental Analytic as an analysis of the a
priori conditions for empirical self-knowledge. According to Guyer,
what Kant calls the fundamental premise of the whole Deduction is the
assumption that I am not immediately acquainted with any manifold of
representations insofar as I think of a representation as contained in a
single moment. Now Kant does say that one must assume in the rest of
his argument that all representations must be in time, since they belong
to inner sense, and are hence subject to synthesis: ˜˜all representations
belong as modi¬cations of the mind to inner sense, and as such all our
cognitions are also subject in the end to the formal condition of inner
sense, namely time, in which they must be ordered, connected, and put
in relations. This is a general remark that one must take as a basis for
what follows™™( ). This is not quite the same thing as taking all
representations to be something that I think at a moment. To be sure,
Kant maintains at   that to think of a manifold as a manifold I must
¬rst represent a sequence of one impression upon another, and he also
says that every representation ˜˜as contained in a single moment™™ can only be
an absolute unity.±° Thus, Guyer is right that Kant thinks that a
momentary representation is not a representation of a manifold. But
° Kant and the demands of self-consciousness
Guyer draws an implication from this assumption of Kant™s that comes
from failure to see that the only kind of representations that Kant
regards as momentary are sensations. Guyer interprets Kant™s state-
ment that intuition o¬ers a manifold, that can never be represented as a
manifold, and as contained in a single representation, without synthesis to
mean that we take a present state and judge it to be a representation of
di¬erent times.
On the other hand, Guyer rightly insists that the temporal order of
what is represented is not created out of some kind of diversity that is
before the mind in some non-temporal manner. This would confuse the
synthesis of recognition by means of which the multiplicity of intuition is
determined to be what it is with the synthesis of apprehension by which
a multiplicity of data is ¬rst given in temporal succession.±± But this
caveat leaves it quite unclear as to what sense we are to give to the idea
of judging or interpreting a present state to be a representation of
di¬erent times. Guyer™s use of judgment and interpretation to explicate
Kant™s account of the role that the synthesis of apprehension plays in
our experience of di¬erent representations, as such, seems to introduce
precisely the synthesis of recognition into his analysis of Kant™s concep-
tion of the synthesis of apprehension that he instructs us to avoid.±
Regardless of the merits of his interpretation of the threefold syn-
thesis of apprehension, reproduction, and recognition, Guyer notes
that recognition involves the possibility of interpretative error and thus
the possibility that one did not actually recognize something that one
thought one had recognized. Guyer is clearly correct to insist that
interpretative errors are possible.± But, surprisingly, Guyer regards a
priori certainty of self-identity as something that rules out the possibil-
ity of interpretative error even though it is introduced by Kant as a
transcendental condition for recognition. But, when Kant maintains
that in self-consciousness we are a priori certain of the self™s numerical
identity, there is no evidence that he wishes to deny the possibility of
error in self-ascriptions, as Guyer alleges. Thus, there is no inconsist-
ency in Kant™s position here. If one identi¬es consciousness of self-
identity a priori with a priori consciousness of one™s empirical identity,
then one cannot indeed allow for the possibility that any of the repre-
sentations which one takes oneself to have had could turn out to be
representations which one did not, in fact, have. But there is no reason
to think that a priori consciousness of self-identity is a priori knowledge
of my particular identity as an empirically knowable individual. While
Kant is interested in a priori knowledge of what makes experience
Introducing apperception
possible, he is loath to argue that we can have a priori knowledge of
empirical facts.
Far from thinking that we cannot be wrong in thinking that a
particular set of states belongs to our personal identity, Kant insists in
the Third Paralogism of Pure Reason that ˜˜the identity of the conscious-
ness of my self in di¬erent times is only a formal condition of my
thoughts and their connection, but does not prove the numerical ident-
ity of my subject, in which despite the logical identity of the I, neverthe-
less a change can have occurred that does not allow its identity to be
sustained™™ ( ). The point that Kant wishes to make is that conscious-
ness of my self-identity does not guarantee that I am an individual who is
actually numerically identical over the time of which I am conscious of
myself as being the same person. To make this point in a plastic way,
Kant suggests the theoretical possibility that an awareness of the past
might be passed from individual to individual in a manner that is
analogous to a series of elastic balls that pass on their motion from one to
another. The ¬nal individual in the series could well have a conscious-
ness of the past histories of all the other individuals in the series, and
believe itself to be a single individual that persisted through the series
even though this would be an illusion.
Guyer realizes that the identity of self-consciousness, as he interprets
it, con¬‚icts with the argument of the Third Paralogism in which Kant
rejects the idea that we somehow have a priori knowledge of our
personal identity. But he regards the tension between his reading of the
Deduction and the text of the Paralogisms as a contradiction in Kant™s
own views.± Indeed, on his reading, Kant is inconsistent even in the
Deduction itself, since the central argument for synthesis from recogni-
tion that Guyer defends is said by Kant to depend on the necessity of
representing one™s numerical identity. The inconsistencies disappear
once one realizes that Kant is concerned with a necessity concerning the
way in which we represent ourselves as experiencers and not with a
claim that we have a priori knowledge of our individual identity over
Guyer™s interpretation of Kant™s idea that we are a priori certain that
we can represent ourselves as self-identical is linked to his reading of the
central task of the Deduction. The Deduction must show that there
must be certain kinds of syntheses that we must perform on experience a
priori in order to show that a priori concepts of the understanding, the
so-called categories, have a legitimate use and one that is restricted to
objects of experience. Guyer argues that it is only if we can impose a
 Kant and the demands of self-consciousness
certain connectedness on any possible representation a priori in virtue
of a putative a priori certainty of self-identity that Kant will have
provided a successful defense of the existence of a priori synthesis from
our consciousness of self-identity. For Kant is supposed to need more
than an analytic claim to the e¬ect that I regard all the representations
that I take to be mine as belonging to my single self, he is supposed to
need a de re necessity that whatever my representations may be they can
be called mine by me. A de re necessity concerning representations would
apply to representations independently of how we happen to pick out or
characterize those representations. Thus, I would know independently
of any conditions governing my recognition of representations that all
my representations are mine. For only this kind of a priori certainty of
self-identity is supposed to require the kind of synthetic principles a
priori governing the connections between representations that Kant is
trying to establish.±µ Since Guyer regards the notion of a priori certainty
of self-identity as ˜˜profoundly questionable,™™ he takes the appeal to a
priori certainty of self-identity to call into question Kant™s general
argument in the Deduction for a priori synthesis from transcendental
But why does Guyer think that a priori synthesis would have to
involve a metaphysical or de re necessity, rather than a purely conceptual
or de dicto necessity? Guyer concedes that Kant sometimes avails himself
of a notion of transcendental synthesis as non-empirical conditions on
experience rather than as acts of transcendental synthesis that are
independent of empirical acts of synthesis: ˜˜There is indeed a transcen-
dental synthesis which however concerns nothing more than the condi-
tions under which the perception of a thing in general can belong to
possible experience™™ ( ·±/ ··).±· Guyer rightly argues that in the
Deduction Kant is committed to the existence of a guarantee that all
representations can be combined in a single self regardless of the
content, and this forces him to appeal to a more robust notion of a priori
synthesis. It is unclear to me why such a guarantee is not consistent with
thinking of a priori synthesis in terms of a priori constraints on empirical
synthesis. But it is undeniable that Kant does think that there is a
guarantee to the e¬ect that any possible object of experience must be a
potential object that can be represented as a representation that can
belong to a possible self-consciousness. And that representation is also
supposed to be represented as a representation that can be compared
and contrasted with other representations of that numerically identical
Introducing apperception
But the possibility, yes, the necessity of the categories depends on the relation
that the whole of sensibility, and with it all appearances, have to original
apperception, in which everything necessarily accords with the conditions of
pervasive unity of self-consciousness, that is, must stand under universal func-
tions of synthesis, namely of synthesis according to concepts, in which apper-
ception can alone prove its pervasive and necessary identity a priori. ( ±±±“±±)

It is to Kant™s argument linking determinate representations to self-
consciousness that I now turn.

 °     ®  © ® ¬  ®  ®  ® ¤    °   ©  © ¬©    ¦
 ¬ ¦-   ®   ©  µ  ®  
Kant draws far-reaching conclusions from his general claim that we can
only make sense of the conceptual component of representations by
reference to a numerically identical point of view that is available in all
experience. He argues that all representations are not only potential
candidates for self-consciousness, but have content and indeed exist only
in relation to a possible self-consciousness:
All representations have a necessary relation to a possible empirical conscious-
ness: for if they did not have this [possible empirical consciousness], and if it
were completely impossible to be conscious of them; then this would say as
much as they would not exist at all. But all empirical consciousness has a
necessary relation to a transcendental consciousness (preceding all particular
experience), namely the consciousness of myself as original apperception. (
Kant attempts to ascribe a di¬erential representational content to
di¬erent representations in virtue of the distinctive representational role
that those representations can play for a possible self-consciousness. But
one obvious problem for his conception is that many representations
seem to have a content that is independent of the cognitive role that they
might play in a possible self-consciousness. It is not even clear that all
representations can become conscious to someone. The possibility of
attributing representations to beings that cannot become conscious of
those representations themselves would seem to be enough to undermine
Kant™s claim that all representations must be potential candidates for
self-consciousness. If animals have representations, but do not have
self-consciousness,then it would seem prima facie that there are represen-
tations of which self-consciousness is not possible. There is no evidence
that Kant thought that non-human animals have self-consciousness. As
 Kant and the demands of self-consciousness
Jonathan Bennett has noted, it would be self-contradictory to demand of a
being de¬ned as non-self-consciousthat it be able to be self-conscious of its

. 1
( 9)