. 4
( 9)


tion or consciousness of the self a priori, representations are ascribed to
the self as subject of thought. Here I represent myself as the formal
subject of thought. I can, however, enrich this formal notion of subject
through introspection and more indirect empirical evidence (included
in the general term ˜˜self-intuition™™). This empirical self-consciousness
di¬ers from self-knowledge. The self is something of which one can and
must have consciousness as a subject, i.e. from the ¬rst-person point of
view. However, the self can only be experienced as an object, something
which has been objecti¬ed, something which has become an object for
consciousness. As an object, the representations of the self are, in
principle, accessible to other points of view. Such experience of the self
as an object of knowledge is based on consciousness of the self as subject
The unity of intuition: completing the B-Deduction
of representation, since the self is only available through the possibility
of a point of view.

  ¬¦ - ®  ©µ ®   ®¤   µ ®© ¦ °    ®¤ ©  
After distinguishing self-knowledge from self-consciousness in section
µ, Kant attempts in section  to establish that even perception must be
subject to the a priori principles governing uni¬cation of representations
in one self-consciousness and hence to the objective standards set by
judgment. In this way, he bridges the gap between his thesis that
perceptual statements are not judgments if they involve an essential
reference to the person having them, and his thesis that self-knowledge
is only possible through non-perspectival statements about inner states.
The key to defending this position is the thesis that all mental states are
also temporal states of human beings and thus have a position in a
shared public time.
According to the received view, the conclusion of the B-Deduction
comes through an appeal to the unity of space and time as a priori facts
of which we have phenomenological evidence through our everyday
experience. The implication of this view is that Kant has no real defense
for the assumption that space and time must be unitary. According to
Allison, the ¬rst step shows ˜˜merely that insofar as unity is introduced
into the manifold of intuition by the understanding, that is, insofar as it
is represented as a manifold, it must conform to the conditions of the
unity of consciousness and, therefore, to the categories. This result
leaves completely unsettled the question of whether data given in
accordance with the forms of sensibility are capable of being uni¬ed in a
single consciousness according to the categories.™™ Thus, the very unity
of self-consciousness would be in jeopardy, if space and time were not
uni¬able in this way. On Allison™s interpretation, the unity of space and
time qua intuitions is somehow independent of the synthesis through
which space and time are said, in the footnote to section  to be given
in the ¬rst place. There is some question as to why the uni¬ability of
intuitions should be an issue at all. Allison does not deny that the
Aesthetic already treats space and time as uni¬able.

But as I argued in Kant™s Transcendental Idealism, we cannot infer the unity of time
(or space) from the unity of consciousness because there is no logical contradic-
tion in the thought of appearances being given in di¬erent times (or spaces).
Consequently, we cannot argue directly from the unity of apperception to the
±° Kant and the demands of self-consciousness
applicability to appearances of the relational (or, indeed, any) categories. We
can, however, reverse the process and argue from the unity of time to the
necessary conditions of the consciousness of this unity. I take this to be the
crucial move in the second part of the Deduction.

Allison™s point must be that objects can be given in the di¬erent
spaces and times of di¬erent experiences. This claim is still ambiguous.
Given the fact that the unity of spatial and temporal intuition, as Kant
understands it, covers the unity of space and time as privately experi-
enced, as publicly experienced, and as physically real, Allison might take
the unity of consciousness to allow for the possibility of di¬erent discon-
nected private, public, or even physical times or spaces. But, given
Allison™s thesis that the unity of consciousness as developed in the ¬rst
step of the B-Deduction is consistent with the existence of objects
existing in di¬erent spaces and times, he must be claiming that a
plurality of either public or private spaces and times is consistent with
the unity of consciousness in either its subjective or objective form.
Contra Allison, it seems to me to be crucial to distinguish the weak
unity of empirical and subjective consciousness from the strong unity of
the consciousness in general that makes such subjective consciousness
possible for Kant. Allison appears to be right that, taken in isolation, the
subjective unity of consciousness is consistent with the disunity of time
or space. However, against Allison, I have argued that the subjective
unity of consciousness is itself parasitic on what Kant calls the objective
unity of consciousness. Thus, if the objective unity of consciousness is
inconsistent with the existence of multiple disconnected phenomenal
times and spaces, then so is the subjective unity of consciousness. Of
course, Allison implies that even the objective unity of consciousness is
consistent with the existence of multiple disconnected times and spaces.
To be sure, Allison does not wish to claim that there are, in fact, multiple
disconnected times and spaces. Instead, he argues that such times and
spaces are ruled out by the unity of our intuition of space and time.
The implication of Allison™s interpretation is that the unity of space
and time a priori is a brute given, or fact about the phenomenology of
our experience that we discover through direct intuition. The unity of
space and time gives our experience and self-consciousness their unity,
and the task of the understanding is merely to represent to itself that
unity. Allison is clearly right that we do experience space and time at
least as if they were each of them necessarily connected in an experien-
tial whole. However, the unity of space and time is, at best, a phenom-
The unity of intuition: completing the B-Deduction
enological fact that needs explanation and defense. The problem with
the Allisonian interpretation is that it o¬ers no prospects for providing
such a defense and thus makes Kant™s claims for the unity and objectiv-
ity of experience ultimately depend on a seemingly ad hoc assumption
that space and time are inherently unitary.
Henrich™s interpretation seems to me to be closer to the mark here.
He argues that Kant is able to include all representations within the
scope of the apperception principle by appeal to the Aesthetic, since
intuition contains all representations in it and now turns out to have
unity due to the synthesis of the understanding.µ Henrich seems to me
to be right to emphasize the manner in which even the unity of intuition
depends for its existence on a synthesis that Kant ascribes to the
spontaneous powers of the mind. Given his thesis that the self-ascriba-
bility thesis needs to be supported by an appeal to intuition, Henrich
needs to argue that this is a new premise. In fact, it is a mere application
of the argument in the ¬rst step of the proof.
By section ±·, Kant already takes himself to have established that
space and time are unitary because they consist for us of representations
that are cognitively signi¬cant for us. These representations are cogni-
tively signi¬cant because they are my representations that are thus
potential candidates for self-consciousness and objective judgment.The
footnote to section ±· tells the reader that the synthetic unity of con-
sciousness is contained in intuitions as singular representations. The
unity of consciousness is characterized as original, thus clearly linking it
with pure apperception. The claim that unity of our intuitions of space
and time is parasitic on the unity of self-consciousness is made explicit in
the summary of the argument in section ±· provided by section ° (
The manifold given in a sensible intuition belongs necessarily under the
original synthetic unity of apperception because [Henrich reads ˜˜weil™™ here as
˜˜insofar as™™] through this unity the unity of intuition is alone possible (section
±·). ( ±)

Here, Kant claims that the unity of intuition is only possible through the
connectability of intuitions in one self-consciousness. This is a stronger
claim than that the unity of intuition (i.e. space and time) is represented
as such only through self-consciousness. It is the claim that unity of
space and time is constituted by self-consciousness. The reference to
section ±· takes up his claim, at section ±·,  ±n, that space and time as
individual representeds have a distinctive synthetic unity of conscious-
±±° Kant and the demands of self-consciousness
ness (where it must be insisted that all such connectedness derives from
an activity of the self ).
In section , Kant appeals to the assumption of the Aesthetic that we
have a priori knowledge of space and time as a whole in order to justify
the claim that all of our perceptions must be empirically connected in a
manner that is compatible with the a priori laws imposed on representa-
tional content by the self (section ,  ±°). These a priori laws are
themselves the laws that make it possible for us to make objective
judgments about spatio-temporal episodes. It is important to note that
Kant insists that the unity of connection involved in space and time ˜˜can
be no other than that of the connection of a manifold of intuition of a
given intuition in general in an original consciousness, according to the
categories, only applied to our sensible intuition™™ ( ±±). There can be little
doubt that this original consciousness is, in fact, the synthetic unity of
impersonal self-consciousness. Thus the unity of space and time is
supposed to be the mere speci¬cation of a relation which holds between
self-consciousness and any intuition to our (spatio-temporal) intuition.
Synthesis is always an activity of the self, and never simply received by
us from objects, as Kant emphasizes at the very beginning of the
Deduction in section ±µ ( ±°). In section ±µ, Kant also notes that the
unity of synthesis is something that precedes all concepts and judgments
( ±±). In section ± it then becomes apparent that the unity that is
higher than all concepts and judgments is the unity of self-consciousness.
But now, in section ,  ±°n, the unity of space and time is supposed to
be a result of synthesis by the understanding preceding all use of
concepts, but displaying itself in perception. How are we to understand
Kant™s claim in section  that the very givenness of space and time as
representeds depends on an activity of synthesis? How can space and
time, which are supposed to be in¬nite wholes existing prior to their
parts, be given through a process of synthesis? The notion of space and
time as in¬nite given wholes suggests a notion of totality that Kant
maintains must escape progressive synthesis by a ¬nite intellect.
Kant can only avoid contradiction by construing the synthesis of
space and time through which space and time are given as in¬nite
wholes as an ongoing process of uni¬cation that never actually comes to
an end. In e¬ect, we must construe the unity of space and time as ideas
of reason, rather than concepts of the understanding. The synthesis in
question must be the pre-conceptual and hence pre-categorial synthesis
of imagination and perception as opposed to the intellectual synthesis of
judgment.· In introducing the distinction between imaginative (percep-
The unity of intuition: completing the B-Deduction
tual) and intellectual synthesis in section , Kant links the perceptual
synthesis of imagination to the ˜˜original synthetic unity of appercep-
tion™™ (section ,  ±µ±). What this means is that imagination must be
guided in its synthesis of perceptual information by the possibility of
uni¬cation of those perceptions in an impersonal self-consciousness. In
this way, it becomes possible for us to conceptualize what we perceive.
At least part of the unity that Kant ascribes to space and time can be
derived from re¬‚ection on what it is to have any consciousness of space
and time as space and time and thus to be able to distinguish space and
time from what is in space and time. The capacity to distinguish the
structure of space and time from objects that occupy spatio-temporal
positions is lacking in subhuman animals. They lack such ability because
they lack the capacity for forming bona ¬de concepts. This capacity is
linked to the possession of dispositional self-consciousness. Self-con-
sciousness allows one to abstract from the current context so that one
can explore alternative possibilities. This ability to conceive of space and
time as abstract structures according to which data may be organized,
may be characteristic of all self-conscious beings. But, even if all ¬nite
self-conscious beings were to represent the world spatio-temporally,
further argument would still be required in order to justify the strong
notion of spatio-temporal unity that Kant assumes. The capacity to
order data spatio-temporally would seem to allow for a plurality of
spatio-temporal orders of things. It does not seem to give the uniqueness
of spatio-temporal order that Kant wants.
The apparent unity of space and time derives from the fact that we
think of di¬erent spaces and times as being connectible in a single
comprehensive point of view. This single comprehensive point of view is
just the impersonal unity of self-consciousness. We need to distinguish
two di¬erent aspects to the unity of space and time corresponding to the
unity of experiences in transcendental and empirical self-consciousness.
In empirical self-consciousness, experiences belong together in my or
your individual mind. Di¬erent individual minds assign di¬erent spatio-
temporal relations to di¬erent experiences. However, these di¬erent
spatio-temporal con¬gurations of experiences are themselves re¬‚ections
of the di¬erent standpoints that di¬erent individuals can assign to each
other within a shared space and time. This shared space and time is
intelligible to di¬erent individuals because they can represent them-
selves as experiencing things in di¬erent ways systematically corre-
sponding to these di¬erent alternative possible spatio-temporal posi-
tions. In doing so, they are forced to abstract from the particular context
±± Kant and the demands of self-consciousness
in which they actually ¬nd themselves and in which they happen to be
experiencing what they are experiencing.
Based on considerations such as these, Kant seems to assume that
there can be only one space and time. But it has been suggested that
experience might, in fact, lead one to believe in multiple space-, or
time-systems. Random occurrences which seemed to violate the unity
of space or time could always be explained away by auxiliary hypotheses
which would not require the extreme measure of giving up on the unity
of space or time. Kant would, in fact, argue that the occurrence of such
events could only be con¬rmed or discon¬rmed against the background
of objects characterized by spatio-temporal continuity. If there were
systematic appearances and disappearances of particulars at regular
intervals, one might, however, be tempted to defend multiple spaces or
times. Systematic appearances and disappearances of certain particulars
could be accounted for by a reformulation of natural laws. Could there,
then, ever be evidence which would lead us to opt for giving up the unity
of space and time instead of reformulating the laws in terms of which we
connect spatio-temporal particulars?
It seems that this is at least a real possibility. The thesis of spatio-
temporal unity seems to rule out singularities in space and time of the
kind postulated by contemporary theories of cosmology. Black hole
physics postulates the existence of quantum tunnelling e¬ects that
provide a form of indirect coupling between disconnected spaces which
gives empirical signi¬cance to disconnected spaces.° Such develop-
ments need not completely dismay the Kantian, since there may well be
a way of accommodating such singularities in a continuous space and
time.± But it is more plausible simply to concede that neither Kant™s
notion of a unity of intuition, nor his notion of the necessary unity of
self-consciousness require that space and time be unique (quasi) individ-
uals unless we accept his claim that space and time cannot be features or
relations of things as they exist in themselves. To concede that claim is,
however, to concede that we cannot legislate to nature except in a
limited sense. We can show that all objects in space and time of which
we can become conscious must belong to a unitary space and time, but
we cannot show that the notion of a non-unitary physical space or time
is inherently incoherent.
° 

Time-consciousness in the Analogies

So far, we have seen that an impersonal consciousness of self can be
regarded as a necessary condition for experience in as much as an
impersonal perspective is built into our ability to interpret the world in
terms of concepts. And, in a very general way, Kant has connected the
possibility of such impersonal self-consciousness with the existence of
categories. The task of this chapter is to explain how the categories can
serve as enabling conditions of experience. Carrying out this task in-
volves an explanation of the link between self-consciousness and the
kind of time-consciousness that is necessary to any experience that is
intelligible to us. For the sake of brevity, I shall restrict my discussion to
the arguments Kant develops in the Analogies of Experience for the
enabling role in experience of the most signi¬cant set of categories: the
relational categories of substance, causation, and interaction.
In contrast to the categories of quantity and quality, the so-called
mathematical categories, Kant does not regard the dynamic categories in
general, or the relational categories in particular, as constitutive of
intuition. Kant insists that there cannot be intuitions that do not have
some kind of extensive magnitude or metric, or some kind of intensive
magnitude, or magnitude corresponding to the intensity of sensation
involved in them. Nevertheless, he does regard the dynamic categories as
constitutive of any concept that we might have of an object in experience:

In the transcendental analytic, we have distinguished amongst the principles of
the understanding, between the dynamic, as merely regulative principles of
intuition, and the mathematical, that are constitutive of the latter. Nevertheless, the
dynamic laws in question are indeed constitutive of experience in that they make
concepts possible a priori, without which no experience would take place. (
/ )

We might be able to have an immediate awareness of the contents of our
perceptual ¬eld, even if the concepts of substance, cause, and interac-
±± Kant and the demands of self-consciousness
tion had no purchase in experience. But, Kant wants to argue, we would
not have any concepts of objects. If it were not a necessary fact about
experience that we are able to apply the concepts of substance, cause,
and interaction to objects of experience, it might turn out that, in fact,
objects of experience could not be identi¬ed and reidenti¬ed across
di¬erent times and spaces. But, if we had no way of identifying and
reidentifying those times and spaces themselves, Kant wants to argue,
we would have no experience at all.
In this chapter, I want to argue that the so-called relational categories
are involved in justifying our judgments about the temporal and less
directly the spatial position of events and things because events present
themselves to us in such a way that we are able to apply the concepts of
substance, causation, and interaction to those events. Unlike commen-
tators such as Melnick and Guyer, I wish to deny that every judgment
concerning the occurrence of an event or a change in a thing involves
the application of these categories to a judgment. Instead, I wish to
argue that judgments concerning the occurrence or non-occurrence of
events or changes involve an implicit commitment to the truth of
principles (metaconceptual judgments), such as ˜˜every event is the
change in the state of a substance™™ or ˜˜every event has a cause™™ in which
such categories ¬gure as concepts. We can make judgments about the
temporal and spatial positions and relations of objects without applying
the concept of substance, cause, or interaction to those objects. But in
order to justify those judgments we need to appeal to the concepts of
substance, cause, and interaction, as well as to the more speci¬c laws
governing substances, causes, and interactions.
The categories in general, and the concepts of substance, cause, and
interaction, in particular, can play a role in justifying our judgments
concerning experiences because perceptions and other inner states are
already given to us in such a way that categories must be applicable to
them. But, in contrast to interpreters such as Allison, I wish to argue that
the categories apply to objective experience because they are indirect
enabling conditions of subjective experience. I wish to reject the idea
that any experience at all must present itself to us in a way that already
involves the actual application of the categories of substance, causation,
and interaction.
Inner states are given to us in time, and have content in virtue of
belonging to a possible self-consciousness. Outer states are given to us
not only temporally, but also spatially. The concepts of substance, cause,
and interaction are required if we are to connect di¬erent inner epi-
Time-consciousness in the Analogies
sodes, especially perceptions, together in a global representation of time
that is capable of supporting an impersonal and standpoint-neutral
representation of self: ˜˜The general principle of the three Analogies rests
on the necessary unity of apperception, in respect of all possible empirical
consciousness, that is, of all perception, at all times™™ ( ±··/ °). In
linking the inner states of di¬erent individuals together in time, the
concepts of substance, cause, and interaction link those inner states to
outer states that are accessible to di¬erent observers. In the process, they
help to constitute a single uni¬ed time and space for all observers.
Now any representational content must be connectable to any other
representational content in one possible encompassing consciousness of
di¬erent representations belonging to di¬erent persons with di¬erent
spatio-temporal positions. For all representations that are intelligible to
us directly represent temporal objects, and at least indirectly, represent
spatial objects, and representations of spatial and temporal objects are
only distinguishable from one another in virtue of the di¬erent contribu-
tions that they make to experience. Thus, the di¬erential contributions
that di¬erent representations make to experience must be su¬cient to
yield a way of distinguishing one space and time from another. Absent
any empirical content to distinguish one space and time from another,
spaces and times may be distinct, in that they have di¬erent relations to
each other, but they are indistinguishable, since there is nothing that
allows one to pick out one term of a spatial or temporal relation from
In the First Analogy, Kant argues that we must postulate substances
as substrates relative to which all change occurs. These permanent
objects with changing accidents make it possible to determine whether a
change has or has not occurred. Substances make it possible for us to
ascertain the truth value of judgments about change by making it
possible for us to set up a time-series in which changes are determinable.
But they make it possible for us to distinguish di¬erent clock-times in a
time-series by providing us with something at those di¬erent times that
allows us to distinguish one clock-time from another.
Even if substances are necessary for setting up a time-series, they are
not su¬cient for ordering times or events. Our knowledge of substances
does not tell us which times or events are earlier, later, and simultaneous
with which other times or events. In the Second Analogy, Kant argues
that causal connections are needed if we are to order episodes in
objectively valid temporal relations of earlier and later. The Third
Analogy then argues that interactions between substances are required
±± Kant and the demands of self-consciousness
in order to establish objectively valid relations of simultaneity. It thus
extends the general analysis of causal relations provided in the Second
Analogy from the temporal relations of earlier. and later to relations of
In accordance with Kant™s claim that subjective experience is para-
sitic on objective experience, the objective relations of earlier, later, and
simultaneous underwrite our ability even to make judgments about a
subjective temporal order to events. It is not that we cannot directly
perceive changes in our mental states or their objects. But, in forming
judgments about even the subjective order of our inner states, we take
on a normative commitment to be able to justify claims about the order
of our inner episodes. Even claims about the subjective order of our
inner episodes can only be sustained by recourse to the way in which the
subjective temporal order of our inner episodes depends on the objec-
tive temporal order of outer episodes. For without a distinction between
my subjective take on what I am experiencing and what I am (subjec-
tively) experiencing, it does not even make sense to say that I am
formulating a judgment about my inner experience. I do not even have
a basis for thinking of myself as having inner experience.
Now, even the relational categories are not enough to elicit an
objective spatio-temporal order from experience; we must also rely on
higher-order principles of reason that guide inquiry in the search for
speci¬c empirical concepts of objects and laws. In the concluding
section of this chapter, I attempt to do this fact justice in a discussion of
the nature of the connection between the general principle that there
are identi¬able substances, causes, and interactions, and the existence
and recognition of speci¬c concepts of substance, causation, and inter-

 µ   ® 
Given the fact that space and time are the forms according to which
anything real must be represented, regardless of whether it is represen-
ted as existing externally or internally to our points of view, any substan-
tive notion of a subject will have to be expressed in terms of the
numerical identity of a spatio-temporal point of view. For, taken in
abstraction from the experience of spatio-temporal particulars, the
concept of substance collapses into the purely logical relation of subject
to predicate.± It seems at ¬rst that the subject in question would also
have to be a spatio-temporal continuant in order to capture the distinc-
Time-consciousness in the Analogies
tion between di¬erent possible temporal points of view. But identity in
point of view across di¬erent temporal experiences does not entail that
the bearer of that identical point of view is the same individual through
those changes in point of view.
Instead of arguing that the self must be a persistent individual over
time, Kant argues that the self must experience persistent individuals in
order for it to be in a position to represent even its identity as a temporal
point of view. The self per se is only a form of experience. The self can
only identify and reidentify an individual across di¬erent spaces and
times if the individual is actually in space and time and hence distin-
guishable from the ¬rst-person point of view that the self must take on all
of its experience. Without locating itself in space and time, there is no
distinction for the self between a true or false judgment about its
persistence across time and space. But, in order to locate itself in space
and time, the self must be able to locate itself relative to other events in
space and time. Kant argues that, in order to be able to locate events
relative to other events, we need to have an experience of an object that
persists over di¬erent events and changes.
Since Kant™s argument for substance, and indeed for causation and
interaction, is limited to the way we must experience the world if we are
to have the kind of self-consciousness that is constitutive of being a ¬nite
rational creature, the argument is limited in its validity to the way
objects must appear to us in experience and hence to phenomenal
(spatio-temporal) substance and its states. The argument for phenom-
enal substance is synthetic. There must be something that one takes to
be the bearer of properties in order for one to have a thought of an
object at all. However, the characterization of the subject of a judgment
as a persistent object is a synthetic claim. It is only relative to the fact of
temporal experience that it makes sense to identify persistence or even
permanence as the criterion for being the kind of object properly
regarded as a bearer of properties. For it is clear that the general notion
of a bearer of properties does not require that the bearer persist over
time. The synthetic character of the argument is emphasized in a note of

Between substance and accidens the logical relation is synthetic. The subject is
itself a predicate (for one can think of everything only through predicates with
the exception of I), but it is consequently only called a subject which is not a
predicate of anything further: ±, since no subject is thought with it; , since it is
the presupposition and substratum of the other. This latter can only be inferred
±± Kant and the demands of self-consciousness
from duration while the other is replaced. Therefore it belongs to the essence
:Wesen9 of a substance that it is persistent. If one supposes that substance
ceases to exist, then the cessation proves that it is not a substance, and since
therefore no substratum is thought of as belonging to this appearance, there are
predicates without a subject, therefore no judgments and no thoughts. (Re¬‚.
µ·, Ak. ©©©, p. ±)

Here Kant argues that the only thing that picks out a substance in
experience is something persistent. What makes the substance an appro-
priate object to be represented by a logical subject, as opposed to a
logical predicate, is that it involves something that persists which then
serves as the real subject or bearer of change. There is nothing in the
meaning of a subject that requires one to think of subjects as things that
persist. Yet, once the notion of substance is interpreted as a spatio-
temporal bearer of properties, then persistence over time is analytic to
the enriched notion of substance that only properly applies to objects of
experience ( ±/ ·).
In arguing that experience requires the notion of a permanent sub-
stance, the First Analogy takes the passage of time that displays itself in
the successiveness of our experiences as a datum. It then attempts to
explain how our experience of passage is possible by appeal to the
existence of permanent substances. The passage of time is marked by
the persistence of temporal order through the shifting successive nows of
apprehension. This allows Kant to argue that time as form of intuition
serves as a persistent substrate for the representation of coexistence and
succession which are determinations of time. Kant insists that time as
the form of experience itself cannot undergo succession ( ±/ ).
Time cannot come to be or pass away, for that would require some
further temporal series relative to which it would make sense to say that
time had come to be or passed away. Insofar as time is the order of
things relative to which change occurs, time cannot be thought to
change on pain of an in¬nite regress. If time were to change, a further
time would always be required relative to which that time could be said
to change. We must think of the temporal order relative to which
change occurs as a tenseless ordering of events according to relations of
earlier, later, and simultaneous. For only such an order is not subject to
succession, since it does not involve a distinction between past, present,
and future.
Since tenseless temporal relations are independent of the shifting
perspective of the present or now, they are not subject to succession.
However, in addition to tenseless temporal relations that are indepen-
Time-consciousness in the Analogies
dent of temporal perspective, there are also inherently perspectival
temporal episodes. There are also the di¬erent nows of apprehension
that replace each other successively. These nows of apprehension are
inherently perspectival, because each state of consciousness picks out its
own distinctive now and that now is the only now that is now for it.
Kant™s initially puzzling remark that time ˜˜as the permanent form of
inner intuition™™ is the ˜˜substrate in which alone simultaneity and
succession can be represented™™ ( ) can be explained. The perma-
nence of temporal order is the way in which the (now independent)
tenseless order of time manifests itself in the successiveness of the
now-series of apprehension.
Kant assumes that we can only perceive objects of experience and
their properties, and not times (or spaces for that matter) themselves.
Times and spaces are only observable by us in terms of the changes in
temporal or spatial position of objects that occupy time and space. From
the fact that time and space are not directly observable, but only
observable through changes in the position of objects in space and time,
Kant concludes that there must be something persistent in the objects of
perception (empirical objects) that allows time (and space) to be repre-
sented, if they are to be represented at all. This persistent something
must survive the successive replacement of the individual nows of
perceptual apprehension. Replacement and coexistence is perceptible
only through the relation of objects of experience to this persistent
object of experience.
We perceive objects. In order, however, to be able to perceive change
or coexistence it must be relative to some perceptible object. Substance
is the persistent substrate of all objects of experience. Substance is not
itself perceptible, but it is that in virtue of which the persistence and
change of perceptible objects can be determined. It is that in those
objects of experience that always remains the same and unchanged.
According to Paul Guyer, there is no compelling reason why perma-
nence in something that is not itself perceptible must be represented by
permanence in something perceptible. After all, Kant distinguishes
between the representation of something permanent and a permanent
representation ( ¬©©©) in the context of articulating his argument
against idealism. The fact that there is no general principle that proper-
ties of what is represented must mirror those of what represents them
would pose a problem were it not for the fact that Kant does not argue
directly from the fact that time is permanent to the permanence of the
substrate that represents it. Instead, Kant argues for the existence of
±° Kant and the demands of self-consciousness
something in objects of experience that allows one to determine whether
two objects of experience coexist or exist one after the other. This also
explains why the detour through the permanence of substance is re-
quired despite the fact that our knowledge that substances are perma-
nent or even persistent in some weaker sense is not immediate, but
inferential. The point is not about the need for a permanent representa-
tion to represent something permanent. It turns on how the relations of
something can be known that is not itself perceptible. Guyer™s second
objection is that Kant shifts from treating substance as something that
serves to represent the permanence of time to something that is the
bearer of properties. But the point of Kant™s argument is that time must
be represented through objects of experience by something that allows
one to determine what the temporal relations between objects are
empirically, and this is whatever it is that counts as the bearer of
If something comes to be, a point in time must precede it in which it
did not exist. Again, if something passes out of existence, it must pass out
of existence at a certain time. To determine that something has come to
be or passed away we must be able to say when such a coming to be or
passing away occurred. We need a procedure for assigning a certain
time to the event immediately before or after a putative coming to be or
passing away. But, if all changes were becomings and passings away, in
other words, if all changes were existence-changes, rather than the
replacement of accidents that themselves belong to persistent substan-
ces, we could not perceive or empirically determine that a change has
occurred at all.
We can assume for the sake of argument that a change is a bona ¬de
case of the coming into being of something out of nothing, or the passing
out of existence of something. What problem arises in this case? If there
is no object that persists through a change, if the change is a genuine
case of coming to be or passing away, that is, a case of something that
comes to be out of nothing or passes away into nothing, rather than a
change in the state of something that continues to persist, it seems to be
impossible to determine whether a change of any kind has occurred at
A basic assumption of the argument is that all changes must be
empirically determinable. The assumption depends on a principle of
empirical signi¬cance that is not universally accepted. It is just not
obvious that in order for there to be a certain change that change must
be empirically knowable. This is an assumption that a metaphysical or
Time-consciousness in the Analogies
transcendental realist will simply reject. Kant thinks that all non-logical
meaning derives from experience, and from this principle he infers the
principle that all cognitively signi¬cant claims must have empirical
signi¬cance, but his ultimate reason for accepting the principle of
empirical signi¬cance is his transcendental idealism. In respect to ap-
pearances, one can show that all changes must be changes in the state of
something that persists, for appearances themselves have existence only
relative to the possibility of being recognized to be thus and such by us.
Such an argument will not go through for things in general. For we
cannot infer with respect to any thing at all that its change must be
observable, unless, as Kant argues, the only things that can undergo
changes are things that undergo changes that are necessarily determin-
able by us.
There is no reason why the changes of things in general must be
within our ken unless the very notion of change is tied to time and the
structure of time turns out to be somehow necessarily mind-dependent,
as Kant argues in the Transcendental Aesthetic. Kant™s assumption that
we need to appeal to the fact that time is a structure of the human mind
is why he thinks that all ˜˜dogmatic™™ attempts to argue for the necessity
that there are substances are doomed to failure ( ±/ ). Absent
transcendental idealist assumptions, there is no contradiction in the
general idea of an unobservable change. It is arguable, however, that
there is a contradiction in thinking of an object of experience as
something that undergoes unobservable changes. And we can defend
the determinability of changes in objects belonging to experience even if
we reject the further Kantian claim that there are no changes at all that
could occur independently of experience.
Even if one grants Kant™s assumption that changes in objects of
experience must be determinable, the argument is still open to an
objection that may seem to be fatal. It proves, at best, that there must be
things that persist through some interval; it does not prove that there are
any things that must endure forever. Most of the spatio-temporal
continuants with which one is familiar in everyday life and even in the
most arcane domains of natural science are particulars of ¬nite duration
that come and go against a background of relatively persistent objects.
Such continuants suggest the idea of permanence to us only because of
their relative longevity. Not only animal bodies, trees, tables, and chairs,
but also mountains, continents, and electrons are only relatively perma-
nent. All of these things undergo change or transformation over time.
Eventually they disappear altogether. This raises the question of
± Kant and the demands of self-consciousness
whether anything lasts forever at all. Relative persistence seems to be
possible without permanence or absolute persistence.
Kant implies that recognizable changes must be construed as changes
in the accidents of substances. But additional argument is needed to
show that some kind of replacement could not be perceived or otherwise
observed to occur in the same perceptual ¬elds as substances (of ¬nite or
in¬nite duration) without that replacement itself being a change in state
of a substance that endures. Strawson makes the prima-facie compelling
objection that changes could be observable against the background of
persistent objects of which they are not themselves the states. Each of the
objects could then persist through some time without persisting through
all of time. This would allow for the possibility of objects coming into
being and passing out of being, while also providing for the possibility
that objects persist over restricted stretches of time. The problem with
Strawson™s objection is that it assumes that we already have some way of
determining which states of things coexist in the same space at the same
time. Appealing to relatively persistent things in the surrounding space
will only help us in determining whether a change has occurred or not if
we already know which states of those things are simultaneous with a
putative change in state of some other thing. But the spatial relations
between things and their states are no more observable independently of
the things that occupy a certain position in space, than are the temporal
relations between things and their states. In order to relate one event to
another event in space we must already be able to link the one space
with the other space at one time or at another time. But we can only do
this if there is something that empirically distinguishes the one space and
time from the other space and time. But empirical content to distinguish
di¬erent spaces and times is only helpful to us to the extent that such
content also links spatial and temporal positions together in such a way
that we can actually distinguish them from each other.
It seems at ¬rst rather easy to dismiss the worry about how to
determine the spatial or temporal position of objects or of their states.
We do, after all, simultaneously perceive things in space. But there will
always be a further question as to whether the things that we observe as
simultaneous are indeed simultaneous and in the position in which we
perceive them. We can only resolve this question if we know what their
objective position in time and space is. But we can only determine that
position if we are able empirically to distinguish one time and space
from another, and relate those individual times and spaces to public
space and time as a whole. If relatively persistent objects come to be and
Time-consciousness in the Analogies
pass away, there will be nothing that distinguishes the spatio-temporal
point at which they come to be or pass away and hence no way of linking
them determinately to other objects that come to be and pass away. This
is a powerful argument so long as one accepts the idea that spaces and
times must be precisely distinguishable. But, plausible as the assumption
is that spaces and times are always empirically precisely distinguishable,
it is an assumption one could reject. And, in fact, it is an assumption
rejected by quantum mechanics. But, even if it is not true that we can
assign a completely determinate spatial or temporal position to all
objects, it does seem plausible to argue that most of the objects that we
experience must be such that we can precisely determine their spatial
and temporal position.
Now Kant does not just wish to argue that spaces and times are
empirically distinguishable, he also wants to claim that empirically
distinguishable spaces and times all belong to one space and time. But if
some substances were to come to be and others were to pass away, then
the empirical unity of space and time would be disrupted. Space and
time would break up into di¬erent parallel and partially, or perhaps
even wholly, disconnected spatial and temporal series. Kant makes the
point explicitly with respect to time:

Substances (in appearance) are the substrates of all time-determination. The
coming to be of some and the passing of others of these [substances in
appearance] would eliminate the sole condition of the empirical unity of time
and the appearances would then relate to two di¬erent times in which existence
would ¬‚ow on; which is absurd. For there is only One time in which all di¬erent
times must be positioned, not contemporaneously, but after each other. (
±“±/ ±“)

We give empirical signi¬cance to temporal sequence by means of
objects that occupy di¬erent times. In coming to be, a new substance
gives rise to a new sequence of events in time. In passing away, it ends a
sequence of temporal events. But, if there is nothing in experience that
allows one to justify a judgment to the e¬ect that one sequence of events
is temporally (and spatially) connected to another sequence of events in
a certain way, then the commitment to be able to justify the judgments
that one makes will not be satis¬able with respect to such temporal
sequences. One will have no reason to think that we are experiencing
objects that belong to the same time at all. We will then have the
apparent ˜˜absurdity™™ that time is not inherently unitary.
The reason Kant seems to think this is absurd is that he is convinced
± Kant and the demands of self-consciousness
that no empirical evidence could be su¬ciently compelling to force us to
give up our ability to make determinate statements about whether an
object existed in a time during which it was not directly observed. Kant
assumes that we would not be able to make out the laws governing
interaction between the two time-systems. This would make it impossi-
ble to con¬rm or discon¬rm claims about objects outside of one™s direct
perceptual ¬eld. Kant seems to be right that changes in the intersubjec-
tively available perceptual objects of our experience must, in principle,
be subject to con¬rmation and discon¬rmation. Such perceptual objects
must therefore be thought of as states of objects that persist throughout
time. This makes it tempting to think of the objects of physical science as
sempiternal particulars. Now, even if the objects of physical science are
sempiternal particulars, such as quantities of energy, we could still make
sense of the notion of di¬erent disconnected physical times, if we could
formulate laws linking physically discontinuous times.
The existence of singularities in space and time would seem to call
into question the strong claims that Kant wishes to make about the unity
of physical space and time. Such singularities are not only generally
accepted to be empirically veri¬able. There is also widespread belief
among astrophysicists and astronomers that we have su¬cient evidence
to warrant the assumption that they do, in fact, exist. But the possibility
of independent time-systems need not undermine the assumption that
there are persistent substances. For it might be argued from Kantian
premises that the empirical knowability of di¬erent time-systems must
itself presuppose the existence of substances that persist through
changes from one time-system to another. This may also be given the
alternative formulation that every empirically signi¬cant event must
have a cause that is, in principle, knowable. Our ability to identify causal
laws operating between di¬erent time-systems, and hence to provide a
su¬ciently rich notion of the relation of our own time-system to another
time-system to warrant its acceptance as an empirical possibility, pre-
supposes something that persists through the change from one time-
system to the other. One will have to give up the assumption of
spatio-temporal continuity. But whatever comes into being in one
time-system will have to be numerically identical with what has passed
out of being in the other time-system. Without this numerical identity,
there is no empirical basis for the assumption that there are two distinct
time-systems. Thus, one might be able to defend Kant™s claim that there
must be an empirical unity to time, without also accepting his claim that
time is a single unique whole. Alternative time-systems depend for their
Time-consciousness in the Analogies
empirical signi¬cance on something that can be taken to be a numeri-
cally identical particular. The particular in question must be both
distinct from self-consciousness and systematically representable in such
a manner that the conditions governing self-ascription of representa-
tions are satis¬ed. Thus, even if the uniqueness of the one space and
time with which we are familiar in our experience does support the
notion of permanent substance, the notion of permanent substance
seems to allow for more recondite versions of spatio-temporal unity than
Kant believed to be possible.
The argument for the persistence of substance involves several dis-
tinct ideas. The ¬rst idea is that every time and every space is distin-
guishable from every other because of the fact that it is occupied by
something that ¬lls it. There is also the idea that these spaces and times
are connected by means of particulars that must be numerically ident-
ical over all time. Kant expresses these ideas in the following passage:
˜˜Now time cannot be perceived on its own. Therefore the substrate that
represents time in general, and on which all change or contempor-
aneousness can be perceived through the relationship of the appearan-
ces to it [the substrate] in apprehension must be encountered in the
objects of perception, that is, the appearances™™ ( µ). In the second
edition of the Critique, Kant tries to link the idea that there are numeri-
cally identical particulars to the conservation of whatever it is that makes
these particulars what they are. He insists that the ˜˜quantum™™ of
whatever serves as the real substrate of change cannot be ˜˜increased or
diminished™™ ( µ).The conclusion that some quantity must be conser-
ved in nature has been repeatedly excoriated for importing a conserva-
tion principle of Newtonian mechanics into a discussion of transcenden-
tal conditions on the possibility of experience.µ While I think that this
criticism misses the mark, his argument for the conservation of the
quantity of matter in the universe must ultimately be regarded as a
In the Critique no argument is articulated for the conservation of the
number of substances; Kant simply concludes that the number (quan-
tum) of substances must remain constant in the universe from the
assumption that substance does not undergo replacement change. How-
ever, an argument which moves from the premise that phenomenal
substance cannot undergo replacement to the conservation of the quan-
tum of that phenomenal substance is to be found in the Metaphysical
Foundations written between the ¬rst and second editions of the Critique.
Presumably, Kant™s re¬‚ection on arguments concerned speci¬cally with
± Kant and the demands of self-consciousness
the conservation of matter motivated his decision to include the new
claim in the revised edition of the First Analogy. Appealing to the
arguments in the Metaphysical Foundations, many interpreters have now
found the argument for the conservation of the quantity of substances
more promising.
Kant maintains that the only thing that can strictly satisfy the demand
for numerical and qualitative identity in experience is material sub-
stance. This is because material substance is the only thing that has
wholly external relations to its parts and to all other things. It is also the
only thing that is subject to quanti¬able and lawlike relations according
to the preface of the Metaphysical Foundations (Ak ©, p. ·±). Kant is
attracted to the idea that in a material substance all di¬erences in
intensive magnitude would reduce to di¬erences in extensive magnitude
and, indeed, to the number of substantial material units of which a thing
is composed. According to the Metaphysical Foundations (Ak. ©, p. µ), if
substance is construed as what is movable in space, then quantity of
substance will be the number of substances in that space. The quantity
of something that is purely spatial will depend only on the number of
parts it has, all of which are related to each other externally and
spatially. But, if these parts are to be real parts, they must be movable
parts, and, if they are movable parts, they may be said to be corporeal
substances. Kant concludes that if substance is conserved, and the
movable in space is substance, then the quantity of the movable in space
must also be preserved. If the quantity of substance (insofar as it is the
movable in space) were either to diminish or increase, then the substan-
ces which make up what is real in space would have either to come to be
or pass away. Such becoming is precluded by the general argument for
the conservation of substance.
Kant thinks that the conservation argument will not go through with
respect to everything that undergoes change. The intensity or magni-
tude of consciousness is not dependent on the quantity of mental
substances of which that consciousness is composed. Consciousness can
increase or decrease in magnitude without the coming to be or passing
away of constituent substances.· After arguing that the increase or
decrease in intensity of consciousness is compatible with the persistence
of substance, he goes on to suggest that consciousness could disappear
altogether through continuous diminution:

It is logically possible for something to exist up to and at a given period of time
and from that moment on not to exist. It ceases to be and begins not to be in the
Time-consciousness in the Analogies
same moment. There is no contradiction in supposing a simple substance to be
annihilated. (Metaphysical Foundations, Ak. ©, p. )

The supposed contrast between the status of the mental and the physical
vis a vis conservation principles raises a worry about whether even
material substance can be conserved. Kant uses his distinction between
extensive and intensive magnitudes to argue that material substances
are conserved while there is no conservation principle for mental sub-
stances. But, given his own premises, the preservation of the quantum of
substance needs to be understood as the preservation of both the
extensive and the intensive quantity of substances rather than of exten-
sive quantity alone.
There is a crippling di¬culty for Kant™s attempt to base a conserva-
tion principle on material atomism. As he formulates the conservation
principle for phenomenal substance, it concerns the quantum of sub-
stance. In the Metaphysical Foundations, he clari¬es this notion of a quan-
tum as a collection of units. The number of units of matter in a
collection of matter would have to remain constant, if the quantum of
substance is to be conserved. In order for it to make sense to say that the
quantum of substance is a function of the number of units which make it
up, one must assume that the number of units is ¬nite. If the number of
units in any quantum of substance is in¬nite, the quantum of any
particular substance could not be a function of the number of units of
which it is composed. For one could take away or add units to that
quantum without a¬ecting the quantum. But Kant is committed to the
idea that space is in¬nitely divisible. He maintains that the in¬nite
divisibility of space entails the in¬nite divisibility of matter. The in¬nite
divisibility of matter and the external nature of all relations between
parts of matter leads to the idea of a potential in¬nity of atoms. This
prevents us from claiming that each material object is composed of a
certain de¬nite number of material atoms and that the number of such
material atoms is conserved in any closed system. There just will not be
a fact of the matter concerning the number of units of which an atom is
Kant is aware of the di¬culty, and opts to measure quantity of matter
by means of quantity of motion (Ak. ©, pp. µ·“µ). Motion and
matter itself are, in turn, traced back to dynamical forces of attraction
and repulsion. The dynamical construction of matter in the Metaphysical
Foundations does indeed provide a way out of this dilemma. Instead of
thinking of conservation in terms of bits of matter, one can move to the
± Kant and the demands of self-consciousness
forces or energy which distinguish matter from space. This allows one to
circumvent problems concerning atomism and the continuum. Some-
thing that is a continuous quantity in one sense can be a discrete
quantity in another sense. This is true of bits of matter. As bits of space
and time, matter is in¬nitely divisible and continuous. As a physical
quantity resulting from the existence of a certain quantity of force,
matter is discrete. It consists of grains with a determinate metric:
In¬nite divisibility refers only to appearance as quantum continuum and is
indivisible from the ¬lling of space; since the basis for the in¬nite divisibility of
space lies in that [¬lling of space]. But as soon as something is assumed to be a
discrete quantity: the number :Menge9 of units is determinate in it; hence
always equal to a number. ( µ·/ µµµ)

Spatio-temporal continuity is consistent with the existence of discrete
atoms endowed with determinate number as physical quantities. But the
argument cannot depend on the fact that matter is an extensive magni-
tude. So, to the extent Kant wants to argue that material substance
rather than mind is the only thing that could be conserved, he must
appeal to the anomalousness of the mental rather than to its lack of
physical extension. In conclusion, we may observe that Kant™s argu-
ment in favor of an Aristotelian conception of substance, as a substrate
of change, is relatively successful, since he has shown that we need
substances to the extent that changes are to be determinate. However,
his own theoretical assumptions raise major reservations about his e¬ort
to privilege matter (in the narrow sense) as substance.

  µ   © ®  ® ¤ °    ° µ  ¬ ©    °  © 
Kant uses the perception of an event as the starting-point for his
discussion of causation. After arguing that we can only determine
whether an event occurs by regarding such events as changes in things
that persist, Kant attempts to establish that we need to appeal to causal
relations in order to be able to determine the objective order of events in
time.We cannot simply read o¬ the objective sequence in which two
phases of an event or set of events occurred from the order of our
perceptions of those events. There are two important reasons for this. (±)
We apprehend things successively ( ±/ ). Succession arises from
the ¬‚ow or elapse of di¬erent states of consciousness, and the changes in
these states may not always re¬‚ect changes in the things observed by
consciousness. The problem is not that we cannot have introspective
Time-consciousness in the Analogies
awareness of succession.±° But we may be mistaken about the order in
which we perceived what we perceived. For what we have is at best a
short-term memory of the order in which we perceived the two states.
We need some criterion for distinguishing a (veridical instance of )
memory from a false memory of the sequence. () Even if the order in
which we judge ourselves to have perceived a state x followed by a state
y is, in fact, the correct order of those representations, we cannot be sure
that the order of our perceptions of states x and y corresponds to the
order in which they actually occurred. The distinction between the
order in which we perceive a set of states and the order in which those
states occurred in the object of which they are states is crucial to the
proper understanding of the argument from the irreversibility of a
certain perceptual order to the existence of an underlying causal cover-
ing law.
When we (veridically) perceive an objective change, or an event, the
sequence of percepts in that perceptual sequence corresponds to the
sequence of changes in its object. Following Van Cleve, I shall refer to
the circumstance in which the order in which we perceive a change
corresponds to the order of the states in the change itself as ˜˜perceptual
isomorphism.™™±± Perceptual isomorphism is analytic to what it means to
perceive an event, where an event is to be understood as an objective
change. Now Kant distinguishes the successive apprehension of an
event (of an objective change in state) from the successive apprehension
of an unchanging object. He uses the apprehension of a ship going
downstream, and of the coexisting parts of a house, as examples of
perceptions of changing and unchanging objects, respectively. In both
of these examples, a change of perceptual states is involved, since we
represent the items successively and our spatio-temporal perspective
undergoes a change. But only in the case of the ship going downstream
is the object in a state of change that is isomorphic to that of our
perceptions. The order of our perceptions is said to be ˜˜tied down™™ in
apprehension in the case of the perception (or rather apprehension) of
an event ( ±/ ·).±
The perception of a(n objective) change di¬ers from a mere (subjec-
tive) change in our perceptions. When I perceive the event of a ship
going downstream, the order of my perceptions cannot be reversed. I
cannot see the ship downstream ¬rst and then perceive the ship up-
stream. This would be the perception of a di¬erent event, i.e. it would be
the perception of a ship going upstream. Irreversibility of the sequence
in which an event is perceived is analytic to the perception of an event in
±° Kant and the demands of self-consciousness
the sense that the order of our perceptions must correspond to the order
of objective change. The ship might go downstream again after going
upstream. It is quite improbable that the ship could return to precisely
the same type of state on any understanding of what it is to be an event.
And, if event identity is tied to the relations of that event to other events,
it will be quite impossible for strict recurrence to occur. For if A“B are
di¬erent then A in the change A B will have to be di¬erent in content
from A in the sequence B“A for they will be involved with other events
that will make them di¬erent. And, even if the two A event-tokens in the
sequence A B A were type identical, the change from A to B and B to A
would nevertheless be di¬erent. The sequence A to B and B to A would
be irreversible with respect to event-tokens, if not with respect to
The irreversibility of sequence that Kant imputes to perceptions of
changes has sometimes been thought to be the basis for his assertion that
those changes in the state of objects are themselves objective.± By
identifying the order of objective sequence with the order in the subjec-
tive sequence of perceptions, he would come into con¬‚ict not only with
Special Relativity, but also with the way the distinction between subjec-
tive and objective sequence operates in Newtonian (and Leibnizian)
mechanics. For even here the motions in terms of which spatial and
temporal intervals are measured are dependent on the frames of refer-
ence that di¬erent observers have. The identi¬cation of subjective and
objective sequence is incompatible even with our experience of such
phenomena as lightning and thunder. Under normal circumstances our
perception of thunder follows our perception of lightning, although the
lightning and thunder actually occur simultaneously. We also perceive
states of stars as simultaneous with us, many of which may have long ago
ceased to exist. Moreover, if a star lies one light-year from us, we will
perceive the light it emits earlier than the light emitted from a star that is
µ light-years away, even when the light from the star that is ¬ve
light-years away was transmitted almost four years earlier than the light
from the nearer star.
Counterexamples against perceptual isomorphism of this kind are
based on the misunderstanding that irreversibility in perception is
supposed to function as the criterion of temporal and causal sequence.
But knowledge of an irreversible order in perception is supposed to
presuppose knowledge of an objective irreversibility of succession. Thus,
in discussing the example of my perceiving a ship going downstream,
Kant notes: ˜˜In our case, I will therefore have to derive the subjective
Time-consciousness in the Analogies
sequence of apprehension from the objective sequence of appearances,
since otherwise the latter is completely indeterminate and does not
distinguish any appearance from any other™™ ( ±/ ). Kant is not
claiming, to be sure, that we cannot have any beliefs about the temporal
order of our perceptions without knowing the objective sequence of the
objects of which they are perceptions. But we can only know that a given
sequence of states in which we perceive something is irreversible, that is,
must occur in a certain order, if we know the object perceived.
We only become aware of changes in our beliefs and desires by virtue
of the way they represent changes in what we represent outside of
ourselves. Once we have some perception of changes in things outside of
us, then we can assign some order to our representations. For, without
the possible perspective of an outside observer to draw on, I have no
sense that I or my representations are really in time at all ( ). From a
purely ¬rst-person perspective, the temporal order of my representa-
tions is indistinguishable from my present take on the temporal order of
those representations. It is not until I see that my present take might be
false, by thinking that another person might assign a di¬erent temporal
order to my representations, that I am really able to form judgments
about the temporal order of my inner episodes. That order will then
only be knowledge properly so-called to the extent that we are able to
derive the subjective ordering of time-sequences from an objective
ordering of changes in spatial objects by reference to the standpoint of
perception. If A is the ¬rst stage of some event and B is the stage which
follows it, then, given the isomorphism de¬nitive of event perception,
there will be perceptual states A and B corresponding to A and B. The
order of the occurrence of A and B is subject to this irreversibility rule
that is de¬nitive of the concept of event perception.
So far, all we know is that if we perceive an event, then the irreversi-
bility rule applies to perceptions. This does not tell us how we know
what the order of change from state A to state B is. Assuming that one is
perceiving an event, the order of what is perceived is ¬xed. The act of
perceiving A will be followed by the act of perceiving B, just as the
event A will be followed by the event B. We have not yet explained,
however, how it is that, given an act of perceiving A followed by an act
of perceiving B, we are then able to determine that an objective change
from A to B has occurred. For it always seems possible that the order of
our representations A . . . N may not correspond to the order of
changes in things A . . . N that are independent of our particular
standpoint. There must be some connection between A and B such that
± Kant and the demands of self-consciousness
we can know that if A occurs then B occurs. Once we know what this
connection is, we are then in a position to determine whether we are
correct in our belief that A precedes B. We can determine whether the
order of A and B according to which we perceive A and B corresponds
to the actual order of A and B. And only if we are in possession of such
facts are we able to challenge even our belief that A preceded B.
Whenever we perceive what appears to be a change in state, we are
presented with opposite states that are perceived sequentially. Percep-
tion alone does not tell us the objective relation between two successive
states of objects perceived, although we are conscious of the fact that the
one state precedes the other in our consciousness. Now, two temporal
states constitute an event only if they are opposites. It thus appears to be
su¬cient to know that they are opposites in order to know that, if they
are parts of an event, they are non-identical parts of an event or
sequence of events. For, if they are opposites, they cannot be simulta-
neous states of the same thing.± If we perceive a state B and we know we
have immediately previously perceived a state A that is opposite to B
and we know that B is at the same spatial location as A, then we know
that A cannot have occurred at the same time as B. If we know that two
states are opposites and states of the same thing, then we know that the
two states must belong to that thing at di¬erent times. And from the
existence of opposite states at di¬erent times we can conclude that a
change has occurred. However, we can only determine which of the two
states is earlier if there is something about the opposite states such that
the one could only exist after the other.
Kant infers that there is a necessary connection between an occur-
rence (an event) and something that precedes it from the rule deter-
minedness of the perceptual sequence in an event perception:

According to such a rule there must therefore lie in that which precedes an
occurrence in general a condition for a rule according to which this occurrence
always and necessarily follows. Thus since it is something that follows, therefore
I must necessarily relate it to something else in general which precedes it and
which it follows according to a rule, i.e. necessarily, so that the occurrence as
that which is conditioned gives certain indication of some condition, and this
determines the occurrence. ( ±/ )

Kant does not directly argue for the claim that objective succession
according to a necessary and strictly general rule is identical with our
notion of causation. One might wonder why succession according to a
necessary and strictly general rule would have to be construed as
Time-consciousness in the Analogies
causation. This is ultimately a matter of the appropriateness of terminol-
ogy. Nominal de¬nitions can be as arbitrary as one likes, but a real
de¬nition must be adequate to the explanatory tasks that a concept is to
ful¬ll. The term ˜˜causation™™ is generally taken to mean that which
necessitates a change from some event-type A to some event-type B.
Objective succession that is determined by a necessary and strictly
general rule gives us the regularity of sequence that is generally required
for a conception of causation. It also expresses the idea that a certain
occurrence generates another occurrence. The one occurrence had to
occur given the assumption that a certain other occurrence preceded it.
What then is the basis for the postulated necessary connection be-
tween events? It is tempting to trace the necessity in question back to the
concept of an event perception. But the analytic principle involved in
perceptual isomorphism, based on the concept of what it is to perceive a
perception, cannot itself provide the basis for the categorical necessity
that Kant must justify.±µ Since he regards causal connection as synthetic
and a priori, he does not intend the mere analysis of the meaning of
event perception to be the key premise in his argument.± How can the
claim that there is a necessary connection between individual tokens of
certain event-types be justi¬ed? Part of Kant™s answer is that the exist-
ence of an earlier time is a su¬cient condition for the existence of a later
time. Time is supposed to have a direction that is represented a priori (
±/ ;  ±/ ). This leads to the idea that it is a necessary law
that earlier times determine later times.
Kant takes the direction of causation to be an expression of temporal
direction. However, temporal direction is not simply causal direction or
reducible to causal direction in any interesting sense. Nor is causal
direction just temporal direction. The notion of cause, based as it is in
the non-material conditional and its ability to support counterfactuals, is
irreducible to the direction of time. In this respect, it must be distin-
guished from the essentially temporal notion of causal direction. This
allows for a conception of causal action that is not essentially temporal in
character, since it does not involve causal direction. Now, despite the
manner in which Kant wishes to link causal connection to the direction
of time, his notion of temporal causation is consistent with the possibility
of the backward causation postulated in some recent physical theories.
He nowhere indicates that the direction of time and the direction of
causation are directly linked in a manner that precludes causal vectors
from traveling from the future to the past. In fact, his theory of action,
which provides the model for his more general conception of teleology,
± Kant and the demands of self-consciousness
involves the idea that action is caused by the agent™s anticipation of a
certain outcome from that action. The agent™s beliefs concerning the
outcome of a certain action gives rise to a desire or aversion in anticipa-
tion of the feeling of pleasure or displeasure that the outcome would
It seems to be a necessary feature of a ¬nite temporal consciousness
that it experience changes in a direction that is irreversible in the order
of succession. But this does not rule out the possibility that the laws or de
facto regularities governing objective change give rise to reversals in the
order of temporal changes. Some objective sequences, such as the
motion of a ship, and perhaps even all objective sequences, may turn out
to be in some sense time-order reversible. The laws of Newtonian
mechanics are, for instance, invariant under time reversal. From the
point of view of the physics of Kant™s time, all processes are, in principle,
reversible. This seems to be true for the fundamental laws of contempor-
ary physics as well. There are phenomenological laws such as the second
law of thermodynamics that are not time-order invariant, but these laws
are based on other statistical laws that are time-order invariant. It seems
quite possible for there to be spatial and temporal regions in which there
is a constant increase in negentropy rather than entropy. This reverse
order of change to our own would not appear to observers in that part of
the universe to be time-order reversed, although our part of the universe
would appear that way to them and theirs to us.
The synthetic a priori necessity that later events follow earlier events
is part of Kant™s reason for thinking that there are necessary (causal)
connections between events. But perhaps a more fundamental reason is
to be found in the idea that changes must be determinable. The
necessity that changes be empirically determinable provides the key
premise in Kant™s argument that every event must have a cause and that
tokens of a certain event-type must be connectable to tokens of some
other event-type according to laws that ¬x the order of their occurrence
in objective time.±· Now, the transcendental or metaphysical realist
simply accepts the possibility that there may be changes that we cannot
explain and that there may therefore be events that have no cause. But,
for Kant, a judgment that a change has or has not occurred must, in
principle, be justi¬able by us. Transcendental idealism not only de-
mands that the subjective conditions under which evidence is available
to individual observers be subject to constraints that apply to all ob-
servers regardless of their standpoint, it also limits empirical judgments
to claims that can, in principle, be supported by evidence. Transcenden-
Time-consciousness in the Analogies
tal idealism limits change to what, in principle, is determinable by us or
at least by a being with ¬nite powers of understanding. This is why Kant
thinks that ˜˜if appearances were things in themselves, then no human
being would be able to ¬gure out how the manifold is connected in the
object from the succession of representations of that manifold™™ ( ±°/
The distinction between the subjective order of representeds and
their objective order amounts, for Kant, to a distinction between contin-
gent and necessary connections between those representeds. Represen-
tations are true of an object, if and only if they cohere ˜˜necessarily™™
amongst themselves. Here the object is just whatever all of our beliefs
correspond to when they are internally coherent. The necessity that the
object transmits to those representeds is itself nothing but the necessary
connectedness of those representations ( ±·/ ). This aspect of the
argument is a direct application of the recognition argument in the
A-Deduction, where Kant ultimately based the possibility of conceptual
recognition on the possibility of an impersonal self-consciousness of
diverse representations. Only such an impersonal self-consciousness is
capable of supporting the normative claim implicit in conceptual recog-
nition that one is representing or rather judging the item in question as
anyone ought to do so.
The idea that what makes for objective relations between items that
we experience is that such items must be connectable in self-conscious-
ness in a way that allows conceptual recognition, has an immediate
relevance to the problem of how temporal episodes may be ordered in
time. The only way of assigning a determinate temporal position to
what is experienced is relative to all other items of experience: ˜˜the
appearances must determine for one another their position in time, and
make their time-order a necessary order™™ ( °°/ µ). A necessary
order is invariant with respect to changes in observers and their spatio-
temporal positions. In this invariant spatial and temporal order, a
position is provided for all possible empirical consciousness. By ˜˜carry-
ing the time-order over into appearances and their existence™™ ( °°/
µ), we get the notion of an object that is independent of any particular
standpoint, and at the same time a sequence of percepts that corre-
sponds to the a priori structure of intuition. We must identify di¬erent
times relative to di¬erent events that occupy them because this is the
only way we have of identifying di¬erent times. ˜˜For only in appearance
can we empirically know this continuity in the connection of times™™ (
±/ ). Our perceptions are true of their objects just in case they are
± Kant and the demands of self-consciousness
connected in a lawlike way to those objects. To order perceived events
in time in a way that corresponds to the objective order of those events,
we must identify the causal laws that connect those events to other
When we perceive an event, we take it to be something that necessar-
ily follows something else. We assume that if we perceive the set of
circumstances that ought to bring about a certain event and the event
does not occur, we are merely imagining or dreaming that the appropri-
ate circumstances are on hand ( °/ ·). The information link
between those perceptions and their object is then deviant. To be sure,
in order to determine that an information link is deviant, we would have
to have knowledge of all of the relevant conditions governing a suc-
cession from one state to another. This is unattainable by us, since the
number of ceteris paribus clauses that may be relevant to the occurrence of
events is potentially in¬nite. We rely on knowledge of the most relevant
conditions. Determining which conditions are most relevant is itself a
matter of skill in judgment, and is inherently interest-relative.
Now, in contrast to the holistic account of time-determination that I
have been defending, Allison insists that the argument in the Second
Analogy is concerned only with the succession of one state to another
that makes up an event, rather than with a sequence of di¬erent
events.± His distinction between the sequence of states in an individual
event and in a series of events tends to be undermined by the nature of
the identity conditions for events. Allison admits that an event is to be
construed as a succession of states. Individual momentary states do not
involve either replacement change or changes in state. They cannot
therefore constitute an event, as Kant understands the notion of event.
And, given the Kantian assumption that time is continuous, there will be
a further momentary state between any two momentary states. This is
an explicit part of Kant™s general defense of the continuous character of
change and causal in¬‚uence: ˜˜Between two moments :Augenblicke9
there is always a time, and between any two states in the two moments
there is always a di¬erence which has magnitude™™ ( °/ µ). Since
all temporally extended states are themselves divisible into further
temporally extended states, there is no fundamental distinction between
an event such as the melting of a piece of ice, or a ship sailing down-
stream, and an event comprising almost the whole time-series.
The example of a ship sailing downstream consists of a sequence of
events, and both the A and B edition formulae for the principle of
Time-consciousness in the Analogies
causation explicitly cover all changes, not just changes from one mo-
mentary state to another ( ±/ ). In the Critique of Judgment, Kant
does maintain that the condition for subsumption (the schema) under
the category of cause is that of ˜˜the succession of the determinations of
one and the same thing™™ (Ak. , p. ±). First we may note that bona ¬de
things are supposed to be substances that last forever. Moreover, Kant
certainly does not think that the relation between cause and e¬ect is
restricted to the states of one thing. Knowing what caused a change
from one state of one thing to another allows us to order those states in
an objective temporal order. However, even though the change is from
one state of the same thing to another, the cause of that change is often
to be sought in some other thing.
Thinking of causation in terms of the successive determinations of
one thing seems to run together the order of causation with the temporal
order of events. However, we do not wish to argue post hoc, ergo propter hoc.
Today may indeed follow yesterday and tomorrow follow today with
necessity. But, as Reid argued against Hume, and Schopenhauer later
argued against Kant, this does not yet establish anything about the
manner in which yesterday™s events are connected to today™s or tomor-
row™s events. One event may precede another event without being its
cause. Kant avoids the fallacious conclusion by arguing that when
something succeeds something else there is some cause of the change
from A to B, but that will have to be sought in some further state, or
rather set of states, C, that precedes and is non-identical with A. In the
example of a ship going downstream, the objective change involves
di¬erent states of a ship and water. Neither the ship nor the water is,
strictly speaking, a physical substance, although they are dependent
particulars supervening on the states of such basic particulars. The
changes which the water and the ship undergo are changes which we
can only understand by reference to causation. However, the ship being
upstream is not the cause of it being downstream, rather this change is to
be understood in terms of underlying causal conditions. This means
that, in a sequence of states which is isomorphic to the perceived order
of those states, the earlier state A cannot generally be said to be the cause
of the later state B. Thus, not every necessary succession from one state
to another state involves a causal connection between those two states.
In most instances, the lawlike change from A to B is itself the e¬ect of a
set of causal conditions whose most prominent member is some event
other than A or B that we can then refer to as an event of type C.
± Kant and the demands of self-consciousness
© ®      © ®
The Third Analogy argues for interaction between all substances as the
basis for objective determination of simultaneity. In Kant™s argument
for interaction, the order-indi¬erence of the perceptual sequence of
contemporaneous states plays a role strictly analogous to that played by
the irreversible order of event perception in the Second Analogy. Again
the strategy is to assume the order-indi¬erence of perception as de¬ni-
tive of simultaneity, but presupposing perceptual isomorphism. Al-
though Kant initially implies that order-indi¬erence is a necessary but
not a su¬cient condition for determining the simultaneity of two states,
it is not really even a necessary condition. Deviant causal chains can give
rise to non-order-indi¬erent sequences of perceptions of simultaneous
states. This complicates the situation somewhat, but does not have a
crucial e¬ect on the main argument. The crucial test of simultaneity is
not the order-indi¬erence of the sequence of our perceptions, but rather
the dynamic interaction upon which our perception is based. Kant
rightly stresses the need for a causal medium to transmit information
from objects to our senses in order to determine relations of simultane-
It is easy to note in our experiences that only continuous in¬‚uences through all
positions of space can guide our sense from one object to the other, that the
light which plays between our eye and the heavenly bodies e¬ects an indirect
community between us and them and through it proves the contemporaneous-
ness of them, that we cannot change any place empirically (cannot perceive this
change) without matter making the perception of our position possible every-
where, and it can only demonstrate its contemporaneousness through its
reciprocal in¬‚uence and through it its coexistence with the most distant objects
(although only indirectly). ( ±/ °)

Schopenhauer argues that there cannot be a play of light between the
eye and the distant star, and hence there cannot be interaction, as Kant
claims. But this ignores the fact that some light will generally be re¬‚ected
back by the eye to its source, although there will be a considerable
time-lag between re¬‚ection and incidence. Schopenhauer also argues
that Kant™s use of light from a distant star as a signal for determining
simultaneity is empirically false, since light takes time to travel from the
star to us.± However, there is no textual evidence that Kant endorses
the empirically false assumption that light travels at an in¬nite velocity.
Kant was in a position to know to a fairly close approximation what the
actual velocity is that light travels at. So the assumption that he thinks
Time-consciousness in the Analogies
that light travels in¬nitely fast depends on the assumption that he
confuses the optical perception of simultaneity with objective simultane-
ity. While we may perceive two events as simultaneous, it does not
follow from the fact that we perceive them at the same time that they
are, in fact, simultaneous. Again Kant was in a position to draw the
distinction in question, although he does not explicitly do so. Schopen-
hauer™s objection is based on the tendentious assumption that Kant
takes there to be instantaneous causal transmission between states across
space, rather than indirect evidence of simultaneity relations on the
basis of states that are already in the causal past of the object to which
they are transmitted. Kant does not explicitly commit himself to the
existence of instantaneous causal transmission across ¬nite distances,
although gravitational force in Newtonian mechanics does depend on
instantaneous action at a distance. Immediate transmission of causal
in¬‚uence between contemporaneous states of substances at di¬erent
spatial locations is consistent with his account of simultaneity in the
Third Analogy. But the Third Analogy does not entail the existence of
unmediated or superluminal causal transmission.
Kant was not in a position to draw all the distinctions with respect to
simultaneity that we would now want drawn. Special Relativity intro-
duces a failure of transitivity with respect to objective relations of
simultaneity that Kant did not anticipate. There is no reason in classical
mechanics to assume that events that are perceived subjectively as
simultaneous with each other must be objectively simultaneous with
each other. However, if an event e± is objectively simultaneous with an
event e and the event e is objectively simultaneous with an event e,
then e± will be identical with e. This no longer holds in Special
Relativity. We cannot simply assume simultaneity to be a transitive
relation. If an event e± is simultaneous with an event e, then this by no
means entails that if e is simultaneous with a third event e, then e± must
also be simultaneous with e. Kant probably assumes that objective
simultaneity relations are transitive. But this does not signi¬cantly a¬ect
his account of simultaneity, since it is concerned with the conditions
under which objective relations of simultaneity may be distinguished
from subjective relations of simultaneity.
Kant does not formulate his notion of interaction and simultaneity as
a relation between events, but as a relation between substances. The
First Analogy requires that substances are sempiternal. Such substances
can interact with each other even if there is no instantaneous action at a
distance or superluminal velocities of transmission. For such substances,
±° Kant and the demands of self-consciousness
simultaneity of states is indirectly de¬nable by appeal to states of the
same substance that are in the causal past or future of those states. His
conception of substances as sempiternal persistents re-establishes the
validity of the assumption that substances observed to be contempor-
aneous with the percipient must also be in causal interaction and hence
contemporaneous with one another. But it is important to note that this
by no means establishes that states of two substances that appear to be
contemporaneous with one another need also be in causal interaction
with one another. Only substances that can neither be created nor
destroyed must interact. Even if such substances are in¬nitely distant
from each other and the forces connecting them can operate only at
¬nite velocities, in the in¬nite length of time during which they exist,
they will eventually interact.
In this chapter, I have looked at the role of substance, causation, and
interaction, in establishing not just an objective, but also even a subjec-
tive order of episodes in time and space. I have argued that some notion
of substance, cause, and interaction is needed if we are to be able to
make judgments about public objects or even to be able to make
judgments about private experiences. These notions turn out to be
conditions under which we can be conscious of distinct spaces and
times, and hence of distinctions between di¬erent representations.
In the next chapter, I want to look at the manner in which we come to
formulate laws governing substances, and their causal interaction. For it
is these laws that Kant takes to underwrite our ability to form empirical
concepts of objects and with them to understand the associations in-
volved in even the most minimal notion of experience.
° ·

Causal laws

In the last chapter, I developed the argument in the Analogies of
Experience for the principle of substance, and for the general causal
principle and principle of interaction. In this chapter, I discuss the
relation of the general causal principle, the general principle of interac-
tion, and the general substance principle to the existence of speci¬c laws
governing causes, interactions, and substances. First I discuss the rela-
tion of speci¬c causal laws to the general causal principle. I argue that
the general causal principle entails the existence of speci¬c causal laws,
but does not entail any particular causal law.
Then I take up the question of the extent to which our knowledge of
speci¬c causal laws depends on a priori knowledge. I argue that Kant
thinks of the causal necessity of particular causal laws as parasitic on a
universality and necessity that cannot be derived from experience. But I
reject the view that Kant wants actually to derive individual causal laws
a priori. I then argue that even probabilistic laws exhibit the kind of
necessity and universality that Kant requires of a causal law.
The existence of probabilistic causal laws governing human action
leaves room for indeterministic causal explanation of human behavior.
However, I argue that Kant insists on the regulative ideal of determinis-
tic causal explanation for human behavior. When this regulative ideal is
taken to be a constitutive principle of experience, a con¬‚ict arises with
the assumption that individuals are capable of free action. Kant resolves
the problem by noting that a complete causal explanation of human
behavior is never actually possible for us even in principle, even though
it is a regulative ideal in our explanation of human behavior. The
¬rst-person perspective of self-consciousness involves a kind of indepen-
dence from causal determination that Kant refers to as spontaneity.
This spontaneity turns out to be only relative to antecedent causes when
taken from the second- and third-person perspective that we must take
in order to observe and understand the behavior of others.
± Kant and the demands of self-consciousness

 µ  ¬ ¬ ·  ®¤    § ®   ¬  µ  ¬ ° © ® © ° ¬ 
The general causal principle defended by the Second Analogy and the
general principle of interaction defended by the Third Analogy tell us
that, wherever there is a change, there must be something that causes
the change to occur. They do not tell us what occurs at the same time as
what, or what follows upon what in objective time. But Kant takes the
existence of the general causal principle and the general principle of
interaction also to involve the existence of a speci¬c set of causal
covering laws that explain the change from tokens of a certain event-
type to tokens of another event-type. Thus, Kant maintains that ˜˜in
conformity with such a rule there must lie in that which precedes an
event the condition of a rule according to which this event invariably
and necessarily follows™™ ( ±/ -). Kant implicitly distinguishes a
¬rst- and a second-order rule. The second-order rule would have to be
the general causal principle that everything has a cause or, rather, that
everything that happens presupposes something upon which it follows
according to a rule. The second necessary and strictly general rule is just
a speci¬c causal law. Given the assumption that the event is repeatable,
Kant is obviously treating the event covered by the law as an event-type
rather than an event-token. The causal principle is thus a condition on
succession that underwrites the existence of universal and necessary
rules governing the occurrence of speci¬c event-types.±
In the last chapter, I argued that, while we can have an immediate
consciousness of a sequence of perceptions, we cannot make a bona ¬de
judgment about the order in which those perceptions occur without it
being, in principle, possible for us to show that a claim that our
perceptions occur in a certain order is correct or incorrect. Some recent
interpreters have made a much stronger claim. Melnick and Guyer
maintain that we cannot even form a judgment to the e¬ect that a
sequence of representations has occurred without knowing the causal
laws to which that sequence is subject, since not even the subjective
order of our representations is ever directly given to us. They appeal to
Kant™s commitment to the inherent successiveness of all of our represen-
tations to justify the ascription of this claim to Kant. However, the
successiveness of representations does not preclude us from having a
direct experience of succession, so long as we take such a direct experi-
ence of succession itself to involve a succession of representations.
Quite apart from the correct interpretation of Kant, it is quite
Causal laws
implausible to claim that, before lawlike correlations in experience had
been discovered, individuals had no awareness of the successiveness of
their experiences. A worry also arises about circularity. In order to form
judgments about succession, we need to know causal laws, but it is hard
to see how one could come to know causal laws without being able to
recognize regularities of succession. There is no evidence that Kant ever
thought that we could only assume that one of our mental states
preceded another if we also thought that we knew all of the relevant
causal laws governing the change from one represented to the other. To
require knowledge of all the relevant causal laws in order to perceive
experiences in a determinate temporal order would quickly lead to
extreme skepticism. Not only would it undermine our ability to ascribe a
temporal order to our states, but also given Kant™s assumptions about
the dependence of content ascription on our ability to assign a temporal
order to our experiences, it would force us to deny that we were able to
self-ascribe inner states, as well as outer states.
A weaker claim seems to be in order. While there is no reason that we


. 4
( 9)