Sie sind auf Seite 1von 22

Philosophy and Phenomenological Research

Vol. LXXVI No. 3, May 2008


 2008 International Phenomenological Society

Whatever begins to be must have


a cause of existence: Humes
Analysis and Kants Response
henry e. allison
University of California, Davis

In T1.3.3. Hume examines the epistemic credentials of the principle


that, whatever begins to exist must have a cause of existence, which
he describes as a general maxim in philosophy (T1.3.3.1; SBN 78).1
Denying that it is either intuitively or demonstrably certain, he concludes that it must be founded on observation and experience
(T1.3.3.9; SBN 82). Even though Hume is careful to indicate that he is
not questioning this principle itself, but merely the [a priori] status commonly assigned to it, he was taken by many of his contemporary critics
to have done just that. Moreover, this apparently contributed more
than a little to the hostile initial reception given to the Treatise and to
Humes well known regrets about having published it. Accordingly, it
is not surprising to nd Hume subsequently insisting that he never
intended to cast doubt on this principle, but merely on its pretensions
to either an intuitive or a demonstrative certainty.2
Although this insistence is often cited as evidence that Hume was not
an arch sceptic regarding causality, this does not entirely resolve the
problem; for the claim that the principle has an empirical origin and the
form of evidence pertaining to propositions of that type is virtually as
1

References to the Treatise are generally given within the text and are rst to the section and paragraph number of the Oxford Philosophical Texts edition, edited by
David Fate Norton and Mary J. Norton, Oxford: Oxford University Press, 2000,
and second to the pagination of the Nidditch revision of the Selby-Bigge Edition
(SBN).

See A Letter from a Gentleman to his Friend in Edinburgh [The letter is included
as an appendix to his edition of An Enquiry Concerning Human Understanding by
Eric Steinberg, Indianapolis: Hackett Publishing Company, 1977,118)]; and Humes
letter to John Stewart of February 1754 (The Letters of David Hume, edited by
J.Y.T. Grieg, Oxford: Clarendon Press, 1969, vol 1, 187).

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

525

controversial as the outright denial of the principle itself.3 Accordingly,


I propose to re-examine Humes argument for his thesis, together with a
contemporary response to it by G.E.M. Anscombe, which is based largely on the rehabilitation of an argument of Hobbes, and the classical
response by Kant. In so doing, I shall try to show (1) that Humes argument has considerable plausibility, if considered within the framework in
which he poses it (his theory of ideas); (2) that while Anscombe points to
some problems in the argument, she fails to locate its major diculty;
and (3) that this is accomplished by Kant in the Second Analogy.
I
To begin with, it is important to realize that Humes discussion of this
principle does not arise naturally in the course of the argument of
T 1.3. The major question at issue there is the nature of probable reasoning, understood in the broad sense to encompass all reasoning
concerning matters of fact; and the fundamental claim is that all such
reasoning is based on the relation of cause and effect. This, in turn,
leads Hume into an investigation of the nature of this relation, which
he maintains involves not only contiguity and succession but also, and
more fundamentally, necessary connection. Proceeding on the basis of
the copy principle, Humes investigation takes the predictable form
of an impression hunt for the latter.4
3

In addition to Kant, the list of critics of Humes attribution of a merely empirical status
to the causal principle prominently includes Thomas Reid, who cites three reasons: 1)
It purports to be a necessary rather than a contingent truth, since it states that every
beginning of existence must have a cause, not merely that it always in fact has one. 2) It
has never been taken to be a generalization from experience (no matter how well conrmed), which, as such, would be open to counter-examples. 3) Experience could never
establish that every change of nature actually has a cause, since for the most part the
causes are unknown and sought for. See Thomas Reid, Essays on the Intellectual Powers of Man, Essay VI; vi (Philosophical Works, edited by Sir William Hamilton, Hildesheim: Georg Olms Verlagsbuchhandlung, 1967, vol. 1,455b-456a). As we shall see
below, the rst and third of these reasons are armed by Kant. Reid diers from Kant,
however, in accepting Humes negative thesis that the causal principle cannot be
grounded a priori and in arming that it does not require such a grounding because it
is self-evident In the contemporary scene, the most vigorous defender of Hume on this
point and responder to the Kantian challenge (as well as to Kantian readings of Hume
in general) is Fred Wilson, Humes Defence of Causal Inference, Toronto, Bualo, London: University of Toronto Press, 1997, esp. 72-84, and. 101-110.

The expression copy principle is the name usually applied to Humes central thesis: that all our simple ideas in their rst appearance are derivd from simple impressions, which correspondent to them, and which they exactly represent (T 1.1.7; SBN
4). As Don Garrett points out, it involves both a resemblance thesis (that every
simple idea has an exactly resembling simple impression) and a causal thesis (that
every simple idea is at least partly caused by a simple impression) (Cognition and
Commitment in Humes Philosophy, New York and Oxford: Oxford University Press,
1997, 41).

526

HENRY E. ALLISON

At rst Hume despairs of nding the requisite impression, which


suggests an outright rejection of the idea of necessary connection. But
rather than arguing in that manner, Hume insists that the latter is an
ineliminable feature of the causal relation and that, if anything, it is
the copy principle that might have to be sacriced.5 Nevertheless, being
extremely reluctant to discard his cherished copy principle, except as a
last resort, Hume decides to abandon his direct survey and endeavors instead to nd some other questions, the examination of which
will perhaps aord a hint, that may serve to clear up the present diculty. Without further ado, he formulates two such questions:
First, For what reason we pronounce it necessary, that every thing
whose existence has a beginning, shoud also have a cause?
Secondly, Why we conclude, that such particular causes must necessarily have such particular eects; and what is the nature of that inference we draw from the one to the other, and the belief we repose in
it? (T1.3.2.14-15; SBN78)

Although the second question does seem directly relevant to Humes


concern, the signicance of the rst is not immediately evident. Moreover, while it is not completely clear what Hume himself thought about
the matter, the two questions seem to be logically independent, since one
might hold that every beginning of existence necessarily has some cause,
while denying that any particular cause must have a particular eect and
vice versa.6 Following L. W. Beck, I shall term the principle at issue in
Humes rst question the every-event-some-cause principle and the
one at issue in the second the same-cause-same-eect principle.7
5

See T1.3.2.12; SBN 77.

This is denied by Wilson, who admits that (1) does not entail (2), but insists that (2)
entails (1) (Humes Defence of Causal Inference, 73-75). I fail to see, however, why
one cannot arm a strict regularity within the causal domain, while maintaining that
there are some events (beginnings of existence) that fall outside its scope. Admittedly,
the second part of Humes fourth rule (the same eect never arises but from the
same cause) might seem to entail the every-event-some cause principle, but it
would only do so if Hume had substituted event for eect. As we shall see below,
Hume himself points out that the claim that every eect has some cause is trivially
true, since cause and eect are relative terms. Nevertheless, the reason why I
remarked that it is not completely clear that Hume himself regarded these principles
as logically independent is because of his sinking(likewise to be discussed below).

Lewis White Beck, A Prussian Hume and a Scottish Kant, Essays on Kant and
Hume, New Haven: Yale University Press, 1978, 120 and passim. Although, given
Humes formulation, one might deem it more accurate to term the latter the particular-cause-particular-eect principle, I believe that Becks formulation better
expresses Humes position. Since the particular causes and eects are viewed as
tokens of a type, it is the assumption that tokens of the one type will be universally
correlated with tokens of the other that is at issue.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

527

Our present concern, however, is with Humes treatment of the former principle. After arguing that it is not founded on knowledge or
any scientic reasoning, and concluding that it must be derived from
observation and experience, he notes that the next question should
be how experience gives rise to such a principle? But rather than
addressing that question, Hume once again abruptly changes course
and states that he nds that,
[I]t will be more convenient to sink this question in the following,
Why we conclude, that such particular causes must necessarily have
such particular effects, and why we form an inference from one to
another?... Twill, perhaps be found in the end, that the same answer
will serve for both questions. (T1.3.3.9; SBN82)

In order to understand Humes procedure here, it is essential to realize


that he is not proposing to sink the rst of his two initial questions into
the second.8 Rather, he is sinking the question of how experience gives
rise to the rst principle into the question of the grounds on which we
accept the second. Hume is, therefore, making a three-fold assumption:
rst, that experience is the source of the rst principle (this is presumably a consequence of the argument of T 1.3.3, which rules out any
a priori source); second, that experience provides the basis for the
acceptance of the second principle, which is precisely what Hume will
endeavor to explain in the subsequent sections; third, that it is by
understanding how experience produces a belief in this principle that
we will come to understand how it produces a belief in the rst as well.
Unfortunately, Hume never explicitly tells us just how his answer to
the second question also answers the rst. But if we keep in mind his
well known view that experience produces a belief in the second principle through custom or habit rather than any process of reasoning, we
may surmise that it is supposed to produce a belief in the rst principle
in the same way. In other words, the development of the custom of
postulating a cause for every beginning of existence must be itself the
result of the constant experience of such beginnings as preceded by
some cause.
If this, or something like it, is Humes view, however, it invites the
question: how do we arrive at the concept of causality in the rst place,
or, alternatively, since Hume is clearly committed to an empirical
explanation of the matter, how does experience teach us that any beginning of existence has a cause? Inasmuch as the possession of the concept seems to be a condition of the possibility of developing the habit
8

528

The contrary is assumed by both Beck, A Prussian Kant, 121, and Wilson,
Humes Defence of Causal Inference, 74.
HENRY E. ALLISON

of associating like causes with like eects and inferring one from the
other, it is dicult to see how the concept could itself be produced by
this habit.9 In short, assuming the logical independence of Humes two
causal principles, it remains mysterious how experience could give rise
to the rst, which, in turn, suggests the need for a fresh look at the
argument of T 1.3.3.
II
Noting that this principle is taken for granted in all our reasoning and
thought to be intuitively certain, Hume initially attacks the latter
assumption by appealing to his theory of philosophical relations.10
Since, according to this theory, the only relations capable of yielding
certainty (either intuitive or demonstrative) are resemblance, proportions in quantity and number, degrees of any quality, and contrariety,
and since the principle in question does not t into any of these categories, Hume summarily dismisses any claim of an intuitive certainty. But
apparently recognizing that this would not impress those who question
his theory of philosophical relations, Hume provides a second argument, which is independent of this theory and intended to show that
this principle is neither intuitively nor demonstrably certain.
Humes argument rests on the premise that the only way to establish
the certainty of a proposition is by showing the inconceivability of its
negation. Consequently, the demonstration in question would require
showing the inconceivability of anything beginning to exist without a

Virtually all that Hume does say that bears on the matter is to be found in his
remarks following the two denitions of cause, where he claims that these denitions support the thesis that the necessity of a cause for every beginning of existence is not founded on any arguments either intuitive or demonstrative. And,
with regard to the rst denition, he notes specically that, we may easily conceive, that there is no absolute nor metaphysical necessity, that every beginning of
existence shoud be attended with such an object (T 1.3.14.34; SBN 172). According to the rst denition, by such an object is to be understood one which is
precedent and contiguous to another, and where all the objects resembling the
former are placd in a like relation of priority and contiguity to those objects, that
resemble the latter (T 1.3.14.30; SBN 170 and 32; SBN 172). As far as I can see,
however, neither of Humes denitions help to explain how experience either gives
rise to the belief or supports the claim that everything that begins to exist is
attended with such an object.

10

For Hume the philosophical relations are basically principles of comparison, on


the basis of which the mind relates ideas for the purpose of cognition. As such they
are contrasted with the natural relations, which are principles of association. Hume
introduces the philosophical relations in T 1.5 and returns to them in T 1.3.1-2,
where he uses them to distinguish between the domains in which genuine knowledge or certainty (either intuitive or demonstrative) is possible and the domain in
which we are limited to probability or opinion.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

529

cause. The argument, which attempts to show that this condition cannot be met, goes as follows:
[A]s all distinct ideas are separable from each other, and the ideas of
cause and effect are evidently distinct, twill be easy for us to conceive
any object to be non-existent this moment, and existent the next, without conjoining to it the distinct idea of a cause or productive principle. The separation, therefore, of the idea of a cause from that of a
beginning of existence, is plainly possible for the imagination; and
consequently the actual separation of these objects is so far possible,
that it implies no contradiction nor absurdity; and is therefore incapable of being refuted by any reasoning from mere ideas; without which
tis impossible to demonstrate the necessity of a cause. (T1.3.3.3; SBN
79-80)

The argument may be broken down into the following seven steps:
1) All distinct ideas are separable.
2) The ideas of cause and eect are distinct.
3) Therefore, it is easy to conceive an object as non-existent at one
moment and existent at the next, without connecting this with
the idea of a cause.
4) Therefore, such a separation (in thought) is possible for the
imagination.
5) Therefore, the actual separation of these objects (in reality) is
so far possible, that it implies no contradiction nor absurdity.
6) Therefore, no reasoning from mere ideas [a priori reasoning]
can negate this possibility.
7) Therefore, it is impossible to demonstrate the opposite.
Before examining the argument, it is important to clarify what
Hume understands by a beginning of existence. Insofar as he refers
to objects, he seems to have in mind the coming into being of
entities, for example, of works of art.11 A more careful consideration,
11

Although Hume is unclear on the point, I assume that he does not mean a coming
into being out of nothing, that is, a creation ex nihilo. In any event, we shall see in
part four that the possibility of a coming to be in this sense (at least as an object
of possible experience) is ruled out by the argument of the First Analogy.

530

HENRY E. ALLISON

however, indicates that Hume is not committed to such a restrictive


view. To begin with, he qualies his claim by denying the possibility
of demonstrating the necessity of a cause for either every new existence, or [my emphases] new modication of existence (T1.3.2.4;
SBN 79). Although Hume does not here tell us what he means by
the latter phrase, in the Abstract he presents as the paradigm case
of the causal relation the famous example of the collision of two billiard balls. Here the motion or impulse of the rst is characterized
as the cause and the motion of the second immediately consequent upon the collision as the eect (AB 9-10; SBN 649-50).
Similarly, in introducing the idea of the causal relation in the Treatise, he remarks that two objects may be considered as placed in
that relation,
As well when one is the cause of any of the actions or motions of the
other, as when the former is the cause of the existence of the latter.
For as that action or motion is nothing but the object itself, considerd in a certain light, and as the object continues the same in all
its different situations, tis easy to imagine how such an inuence of
objects upon one another may connect them in the imagination.
(T1.1.4.4; SBN 12)

It seems clear from this that the category of effect encompasses actions
and motions (both of which may be regarded as new modications of
existence) as well as new existences. In fact, even this is too narrow,
since Hume presumably would be prepared to recognize as effects the
cessation of actions and motions as well as their inception, and, more
generally, any change of state, wherein the object continues the
same: for example, the change of water from a liquid to a solid state.12
Accordingly, I shall take beginning of existence to refer to any
change of state and Humes argument as a challenge to the a priori

12

Humes acknowledgement that in such cases the object continues the same certainly stands in some tension with his later account in T 1.4.2 of the ideas of continued and distinct existence as ctions produced by the imagination, but I cannot
deal with that issue here. We shall see in part four of this paper, however, that
Kant viewed this thesis as established in the First Analogy under the guise of
the All change is alteration principle and that it is central to Kants analysis of
causality.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

531

status of the principle that every such change, that is, every event, has
some cause.13
The rst step in Humes argument is a restatement of his separability principle, which, in its canonical formulation, maintains that,
[W]hatever objects are different are distinguishable, and that whatever
objects are distinguishable are separable by the thought and imagination (T1.1.7.3; SBN 18).14 Although the matter is highly controversial,
I shall treat the principle itself as non-problematic and consider only its
applicability to the ideas of a beginning of existence and of this beginning as having a cause. Moreover, since steps 6 and 7 merely spell out
the negative implications of Humes argument for the project of demonstrating the causal principle, I shall limit my analysis to steps 2-5.
As initially formulated, step 2 is quite misleading; for in response to
those who argue that every effect must have a cause, because tis
implied in the very idea of effect, Hume points out that cause and
effect are correlative terms (like husband and wife) and that the
true state of the question is not whether every effect must have a
cause, but whether every object, which begins to exist, must owe its
existence to a cause (T1.3.3.8; SBN 82). Nevertheless, as this reformulation indicates, the initial difculty is easily remedied. All that is
required is the replacement of step 2 with step 2: For any object or
state O, the idea of O beginning to exist is distinct from the idea of this
beginning having a cause.
Steps 3-5 argue for the separability of these two ideas: rst in thought
and the imagination (steps 3 and 4) and then in reality (step 5).

13

Since we shall also see in part four that this is equivalent to what Kant understands
by an event, it makes it possible to regard Hume and Kant as arguing against each
other on the status of the every-event-some-cause principle, rather than merely talking past one another. Nevertheless, it must be noted that, at least in the Treatise,
Hume uses the term event in a very broad sense. For example, he uses it not only
to refer to occurrences such as Caesar being killed (T 1.3.4.2; SBN 83), and outcomes with various degrees of probability, such as a die landing on one particular
side (T 1.3.11.7; SBN 126) or a ship returning safely to port (T 1.3.12.8; SBN 134),
but also to situations like the experience of a virgin on her bridal night (T 2.3.9.
29; SBN 447), and historical events such as the revolution of 1688 (T 3.2.10.16-17;
SBN 564-65). In the Enquiry, however, Hume tends to use the term in a sense closer to the Kantian to refer to occurrences like the motion of the second billiard ball
after being struck by the rst and, more generally, to causes and eects (See EHU
4. 8-11; SBN 28-30).

14

I say restatement of this principle because Hume here refers to ideas, whereas what
I have termed the canonical formulation cited above refers to objects. Nevertheless,
in his treatment of distinctions of reason Hume apparently equates this principle
with the proposition that, [A]ll ideas, which are dierent are separable
(T1.1.7.17; SBN 24); and this seems to be equivalent to the proposition currently
under consideration. On this point see also the Appendix 12; SBN 634 in connection with Humes discussion of personal identity.

532

HENRY E. ALLISON

The contribution of step 3 is not immediately apparent, however, since


it seems to claim merely that for any object or state O, it is possible to
think of O as coming into existence without also thinking of this
occurrence as having a cause (without conjoining it to the distinct
idea of a cause...). But, as Anscombe points out, while this relatively
innocuous thesis follows non-problematically from the preceding steps,
it is of little help in advancing the argument.15 Rather, what Hume
needs is the substantive thesis (30 ): For any object or state O, O can be
conceived not to exist at t1 and to exist at t2, without this coming into
being having a cause.
A brief glance at Humes account of so-called distinctions of reason, such as that between the shape and color of a globe, may help to
highlight the salient difference between 3 and 30 .16 Although Hume
denies on the basis of his theory of ideas, according to which all ideas
are particular and fully determinate, that one can form an idea of the
shape of a globe without its color or vice versa, he also points out that
one can consider one of these properties without considering the other.
Such a mode of consideration involves a distinction of reason and it
plays a central role in Humes account of how ideas, though in themselves particular, can function as universals. Applying this to our present concern, since it arms merely the possibility of thinking of
something beginning to exist without also thinking of this beginning as
having a cause, it leaves open the possibility that the distinction
between these two thoughts amounts merely to a distinction of reason.
But since in that event Humes conceivability argument would fail, it is
incumbent upon him to show that the distinction is not of that nature.
And for that Hume needs 30 .
Assuming that this is what Hume had in mind puts us in position to
understand the function of step 4, which turns on the identication of
conceiving and imagining. Given this identication, if I can imagine x
without also imagining y, then y does not form part of the content of
my idea of x. If it did, I could not even imagine x without also imagining y, just as I cannot imagine a triangle without imagining a gure
with three sides. But if, as is now being supposed, y does not form part
of this content, then I can not only imagine x without imagining y,
I can also imagine (and, therefore, conceive) a state of affairs in which
x exists and y does not.
15

See G.E.M. Anscombe (Whatever has a Beginning of Existence must have a


Cause, Collected Philosophical Papers, Minneapolis: University of Minnesota
Press, 1981, vol. 1, 96).

16

For Humes account of distinctions of reason, see T1.1.7. 17-18; SBN 25. The relevance of Humes account of distinctions of reason to our issue is suggested by
Barry Stroud, Hume, London: Routledge & Kegan Paul, 1977, 49.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

533

In addition to the dubious equation of conceivability and imaginability, which infects Humes epistemology as a whole, there is a localized objection to the present argument. It has likewise been raised by
Anscombe and takes the form of a dilemma: either Hume is proceeding
from the plausible premise that we can readily imagine something
beginning to exist apart from any particular cause to the conclusion
that we can imagine it beginning to exist without any cause at all, in
which case it involves an apparent slide; or he is claiming straight out
that we can conceive something beginning to exist without any cause at
all, in which case it amounts to a sheer assertion.17
In order to address this objection, it is necessary to explore a bit
more fully the idea of something coming into existence from a
Humean point of view. According to Humes theory of ideas, this
may be analyzed as a complex idea composed of the ideas of an
object in two successive and contrary states (e.g., a billiard ball at
rest at t1 and in motion at t2). Such an idea is complex inasmuch as
it contains a combination of the ideas of two distinct states of affairs,
and it supposedly corresponds to a complex impression with the same
content. What is particularly noteworthy, however, is that the idea of
having a cause does not form part of the content of this idea, from
which it follows on Humean principles that the latter can be conceived imagined apart from the former. Accordingly, given Humes
theory of ideas, it is at least arguable that the stronger conclusion is
warranted.
Finally, step 5 involves the move from separate conceivability to the
possibility of separate existence.18 Once again Anscombe objects, characterizing Humes reasoning as based on the Parmenidean principle
that It is the same thing that can be thought and can be. And while
not questioning this principle itself, Anscombe rejects Humes extension
of it to the case of the thought of something beginning to exist on the
grounds that I can imagine or think of a sprig of leaves as existing
without there being any denite number of leaves that I think of it as
having...[though] this does not mean that I can think of it as existing
without having a denite number of leaves.19
Although the latter point cannot be gainsaid, its applicability to
Humes reasoning is questionable. For if the above account of the
17

Anscombe, Whatever has a Beginning of Existence must have a Cause, 97-98.

18

It is not completely clear what sense of possibility Hume has in mind here, since
he qualies the claim by stating that separate existence is so far possible, that it
implies no contradiction or absurdity. What seems to be crucial, however, is that
the sense of possibility be strong enough to block any purported demonstration of
the contrary.

19

Anscombe, Whatever has a Beginning of Existence must have a Cause, 99.

534

HENRY E. ALLISON

complex idea of a beginning of existence is correct, then imagining such


a beginning without a cause is more like imagining the trunk of a body
without the limbs than imagining a sprig of leaves as existing without
having a denite number.20 But since this is a crucial point, it may
prove instructive to consider briey Humes critique of Hobbess
attempted demonstration of what amounts to the contrary thesis,
together with Anscombes endeavor to rehabilitate the Hobbesian argument in response to Humes critique.
III
Hobbes argument is one of three purported demonstrations of the
principle that every beginning of existence must have a cause to which
Hume refers, all of which he dismisses as question begging. In the case
of the other two (those of Clarke and Locke) this is readily apparent,
since the former asserts that if something were lacking a cause it would
have to cause itself; while the latter postulates that if something were
to come into being without a cause, it would have nothing as its cause.
Humes quick dismissal of Hobbes on the same grounds is, however,
somewhat more problematic. According to Humes formulation,
Hobbess argument goes as follows:
All the points of time and place...in which we can suppose any object
to begin to exist, are themselves equal; and unless there be some
cause, which is peculiar to one time and to one place, and which by
that means determines and xes the existence, it must remain in eternal suspence; and the object can never begin to be, for want of something to x its beginning.21

20

The example is from Berkeley, The Principles of Human Knowledge 5 (See also
Introduction to the Principles 10).

21

To cite Hobbes own words: that a man cannot imagine anything to begin without
a cause, can no other way be made known, but by trying how he can imagine it;
but if he try, he shall nd as much reason, if there be no cause of the thing, to conceive it should begin at one time as another, that he hath equal reason to think it
should begin at all times, which is impossible, and therefore he must think that
there was some special cause why it began then, rather than sooner or later; or else
that it began never, but was eternal (Of Liberty and Necessity, The English Works
of Thomas Hobbes, edited by Sir William Molesworth, London: John Bohn, 1840,
vol. 4, 276). Humes gloss diers from Hobbess actual argument in at least two
notable respects: 1) Hume adds a reference to place that is lacking in Hobbes; 2)
whereas Hobbes suggest that without assuming a cause xing the beginning of the
existence at some determinate time, we would have to think that the object existed
eternally, which presumably is contrary to the hypothesis, Hume takes the argument to be maintaining that without assuming a cause determining when and where
it came into being, we would have to admit that it would never exist at all. The
latter brings it closer to the line of argument that Hume attributes to Clarke and
Locke and makes it more obviously question begging.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

535

To which Hume replies:


Is there any more difculty in supposing the time and place to be xd
without a cause, than to suppose the existence to be determined in
that manner? The rst question that occurs on this subject is always,
whether the object shall exist or not: The next, when and where it shall
begin to exist. If the removal of a cause be intuitively absurd in the
one case, it must be so in the other: And if that absurdity is not clear
without a proof in the one case, it will equally require one in the
other. The absurdity, then, of the one supposition can never be a
proof of that of the other; since they are both upon the same footing,
and must stand or fall by the same reasoning. (T1.3.3.4; SBN 80)

Basically, Hume seems to be claiming that, in arguing for the absurdity


of something beginning to exist without a cause simpliciter by appealing to the absurdity of such a beginning at a particular time and place,
Hobbes is begging the question because the two scenarios are on equal
logical footing. If the former can be conceived without any absurdity,
then so can the latter. Conversely, if the absurdity of the former
requires proof, then so does that of the latter. In either event, Hobbes
cannot help himself to the presumed absurdity of the latter in order to
establish that of the former.
Humes curt dismissal of this argument has been criticized, however,
by Anscombe, whose critique has two components: 1) a focus on the
notion of something coming into existence, which largely ignores
Humes underlying theory of ideas; and 2) the drawing of a sharp distinction between imagining and really imagining, that is, seriously supposing or truly judging, that something came into existence without
a cause.22 The key point is that such a supposition requires a consideration of the attendant circumstances of the occurrence, specically its
when and where, which is in accord with Hobbes view but in direct
conict with Humes claim that these are irrelevant to the question at
issue.
Anscombes main argument appears to turn on the distinction
between it coming about that something is present at a certain place
at a given time and something actually coming into existence at that
place (and time).23 Even though they both fall under the description
an object being at placea at t1, which was not there previously,
these two scenarios are not equivalent; for in order for something to
be at a certain place at a given time it need only to have traveled
there from elsewhere, not to have come into existence tout court.
22

Anscombe, Times, Beginnings and Causes, Collected Philosophical Papers, vol. 2,


160.

23

Ibid., 161.

536

HENRY E. ALLISON

Thus, the question for Anscombe becomes how one distinguishes


between these scenarios. Her point is that if one is seriously to suppose that something has in fact come into existence at a certain place,
one must be able to eliminate other possibilities, particularly the possibility that it somehow migrated there from elsewhere. Anscombes
claim is that this determination requires an appeal to a cause. More
specically, in the case of familiar objects such as chairs and babies it
requires understanding the causal process by which entities of that
sort are normally produced.24
Anscombe is correct to insist on the importance of distinguishing
between merely imagining and seriously supposing something coming
into existence without a cause, as well as on the need to consider the
attendant circumstances in attempting the latter. Nevertheless, I nd
her account problematic because of her apparent assumption that,
despite Humes careful qualications, beginning of existence refers
primarily (if not exclusively) to the becoming of whole entities rather
than to changes of state of such entities, for example, the change of a
billiard ball from a state of rest to motion.25 At least her argument,
as I understand it, addresses only the former scenario. To be sure, the
latter may also be described as the coming into being of a new
object (motion), but that seems to be a highly articial way of
describing the situation. Moreover, in such cases the role of causality
is not to assure (or judge) that some object has actually come into
existence at a certain place, as opposed to having traveled there from
elsewhere; it is rather to determine whether a change of state, which
might, but need not, include a change of place, has occurred at all.
And in order to see why this requires the introduction of causality we
must turn to Kant.
IV
In spite of Kants basic opposition to Hume, they are in complete
agreement on the fundamental point that the causal principle cannot
be demonstrated by conceptual analysis and, therefore, not by anything
like the conceivability argument that Hume attacks. As Kant puts it in
a passage, which both refers back to his own transcendental proof of
this principle in the Second Analogy and seems to allude directly to
Hume,
24

Ibid., 161-162.

25

Like Hume, Anscombe is unclear about how she understands the coming into
being of entities; but again I am assuming that she does not have in mind a creation ex nihilo. Certainly, that is not suggested by her examples, e.g., chairs and
babies.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

537

The proof does not show...that the given concept (e.g., of that which
happens) leads directly to another concept (that of a cause); for such
a transition would be a leap for which nothing could be held
responsible; rather it shows that experience itself, hence the object of
experience, would be impossible without such a connection.
(A783 B811)26

This passage also indicates that Kants response to Humes challenge to


the a priori status of the concept and principle of causality is indirect,
amounting to the introduction of a radically new alternative, namely,
that they make possible the very experience from which Hume claims
they are derived. This is the thesis of the Second Analogy, which, in its
second edition formulation, states that All alterations occur in accordance with the law of the connection of cause and eect (B 232).
As this formulation indicates, the Second Analogy presupposes the
First, which maintains that all change [Wechsel] must be understood as
an alteration [Veranderung] of something that persists.27 Viewing the
two Analogies together, they can be seen as addressing two distinct
senses in which something might be thought simply to come into existence, which are glossed over in Humes phrase beginning of existence. One sense is expressed in the traditional theological notion of a
creation ex nihilo, that is, a becoming that is absolute rather than
merely of a new state of an entity that already exists in some form.
This is rejected in the First Analogy, which by arguing that all change
is alteration denies such a possibility. Kant does not, of course, argue
on metaphysical grounds that such an occurrence is absolutely impossible, but merely that it is experientially impossible, inasmuch as it violates the conditions of possible experience. The second sense is that an
alteration or, in Humes terms, a modication of existence, might
occur without a cause. This is the concern of Second Analogy, the
argument for which, insofar as it concerns us here, can be broken

26

References to the Critique of Pure Reason are to the standard A and B pagination
of the rst and second editions. Citations are taken from the translation of Paul
Guyer and Allen Wood, The Cambridge Edition of the Works of Immanuel Kant,
Cambridge: Cambridge University Press, 1998. The reference to the Prolegomena is
to the pagination of volume 4 of Kants gesammelte Schriften, edited by the Koniglichen Preusslichen Akademie von Wissenschaften, Berlin 1911, and the citation is
from the translation by Garry Hateld in Theoretical Philosophy after 1781, The
Cambridge Edition of the Works of Immanuel Kant, edited by Henry Allison and
Peter Heath, Cambridge: Cambridge University Press, 2002.

27

I discuss the argument of the First Analogy in Kants Transcendental Idealism, An


Interpretation and Defense, Revised and Enlarged Edition, New Haven: Yale University Press, 2004, 236-46.

538

HENRY E. ALLISON

down into ve steps.28 I shall rst sketch these steps and then consider
the adequacy of the argument as a response to Hume.
1) Since an event or happening consists in an alteration, its cognition requires two successive perceptions of the object. Consequently, unless I can contrast the present state of the object
with its preceding state I cannot be aware of an event.
2) But since every perception follows upon a succeeding one (all
apprehension is successive) this is merely a necessary and not
also a sucient condition of such awareness. In particular, it
does not provide a basis for distinguishing between successive
perceptions of a static state of aairs (e.g., a house) and successive perceptions of successive states of an object (e.g., a ship
sailing downstream).
3) Drawing this distinction requires an interpretation of successive perceptions as perceptions of successive states and this
interpretation appeals to the notion of irreversibility. In the case
of event-perception we regard the order in which perceptions
are apprehended as irreversible (A-B and not B-A); whereas in
the successive perceptions of a static state of aairs, we regard
this order as indierent.29
28

What follows is a highly condensed version of the interpretation of the basic argument of the Second Analogy, which I provide in Kants Transcendental Idealism,
249-52. I say insofar as it concerns us here, because Kants full argument contains
an additional step in which he claims that not merely the perceptions of events but
the events themselves, qua objects of possible experience, are necessarily subject to
the causal principle. Although this step, which turns on the Kantian principle that
The conditions of the possibility of experience in general are at the same time conditions of the possibility of the objects of experience (A158 B197), is obviously essential to the full argument, I shall not consider it here because it raises thorny
questions regarding transcendental idealism that would take us too far a eld.

29

This appeal to irreversibility is probably the most contentious feature of Kants


argument and a good deal rides on its interpretation. Perhaps the most natural
reading is to take it as a phenomenologically accessible feature of our perceptions,
which one supposedly inspects in order to determine whether or not they yield the
perception of an event. As many have pointed out, however, this does not issue in
a happy result. On the one hand, every succession of perceptions in consciousness,
as a particular succession, is unique and, therefore, irreversible; while, on the other,
irreversibility cannot be used in a non-question begging way as a criterion of event
perception, since it is only if one already assumes that one is perceiving an event
that one can deem the order of perceptions as irreversible in the appropriate sense.
Accordingly, I have argued elsewhere that rather than regarding it as an intrinsic
property of a sequence of perceptions, irreversibility should be taken as characterizing the way in which one connects successive perceptions in thought (the objective
unity of apperception), insofar as one purports to represent through them an objective succession. See Kants Transcendental Idealism, 250-52.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

539

4) This irreversibility results from subjecting the succession of perceptions to an a priori rule, which species how any cognizer
confronted with this sequence ought to construe it. Since the
order is temporal, the rule must be provided by a transcendental
schema, and since it is a rule for the thought of an objective
succession (as contrasted with duration or co-existence) it can
only be what Kant terms the schema of causality, namely, the
succession of the manifold [of perceptions] insofar as it is subject to a rule (A144 B183).30
5) Consequently, to think a succession of states as objective just is
to subsume it under this rule, which entails that one cannot
think of something happening (an event) without also thinking
of it as preceded by some cause.
Assuming that something like the above constitutes Kants core argument in the Second Analogy and granting that a good deal of work by
way of lling in the details and clarifying its underlying assumptions
would be required to make it persuasive, our nal question is whether
such an argument, appropriately eshed out, constitutes a viable
response to Hume regarding causality.31 As a rst step in addressing

30

For an argument in support of the latter claim regarding the schema of causality,
see Kants Transcendental Idealism, 223-24.

31

Recently, Eric Watkins has challenged the premise underlying this often asked
question by maintaining that in the Second Analogy Kant did not even attempt
to refute Humes views, but instead introduced a competing and supposedly superior model of causality (Kant and the Metaphysics of Causality, Cambridge: Cambridge University Press, 2005, esp. 381-390). Watkins oers four reasons for this
novel and interesting thesis, the last and allegedly strongest of which is that the
model of causality that Kant provides in the Second and Third Analogies diers
so radically from Humes that there is no common ground from which a refutation could be mounted. Although there is a good deal of truth in what Watkins
says about the radical dierences in their respective models of causality, particularly if, as Watkins does, one focuses on the Third Analogy, I think that he dramatically overstates his case. First, Kants language, particularly in the statement
of the principle of the Second Analogy in the rst edition and in the passage
cited above from the Doctrine of Method about transcendental proof certainly
suggest not only that Kant had Hume in mind, but that he was attempting to
provide a competing proof. Second, if one focuses on the rst of Humes two
questions, it is clear that, whatever the dierences in their models of causality,
they both addressed the same question, namely, whether the concept and principle of causality are a priori. Hume gave a negative and Kant a positive answer
to this question and this constitutes the heart of their dispute. Finally, by Watkins criterion it seems that there would be very few refutations in philosophy,
beyond those that point to an inconsistency in an opponents position; indeed,
similar claims could also be made about Kants refutations of Leibniz, Descartes, and virtually every philosopher whom he criticizes.

540

HENRY E. ALLISON

this question, it is essential to underscore the limited scope of Kants


concern in the Second Analogy. On my reading, Kant is there concerned only with the rst of Humes two logically distinct questions
regarding causality, that is, with the a priori status of the concept and
principle.32 This is not to say that Kant had no interest in Humes second question, which concerns the principle that like causes have like
eects. It is rather that the latter raises a whole set of dierent issues
and that his response to it is to be found elsewhere, namely, in the
Appendix to the Dialectic of the rst Critique and the Introductions to
the third.33
Limiting ourselves, then, to the rst of Humes questions, the essence
of Kants response is contained in the nal step of the argument
sketched above; for it is here that Kant challenges the separability thesis on which Humes conceivability argument turns. If, as Kant maintains, one cannot cognize or experience (these here being regarded as
roughly equivalent) an event without thinking (or presupposing) that it
has some cause (though not any cause in particular), then one cannot
separate the thought of something happening from the thought of its
having a cause, which amounts to a direct denial of the conclusion of
Humes argument.

32

I believe that this reading is supported both by the formulation of the principle
of the Analogy, particularly in the A-version, where it states that Everything
that happens (begins to be) presupposes something which it follows in accordance
with a rule (A 189): and by the structure of the argument as I have here interpreted it. I further believe that it better accords with a key passage, in which
Kant remarks about his conclusion [that the understanding imposes the causal
principle upon appearances as a condition of the possibility of the experience of
something happening] that it seems as if this contradicts everything that has
always been said about the course of the use of our understanding, according to
which it is only through the perception and comparison of sequences of many
occurrences on preceding appearances that we are led to discover a rule, in accordance with which certain occurrences always follow certain appearances, and are
thereby rst prompted to form the concept of cause. On such a footing this concept would be merely empirical, and the rule that it supplies, that everything that
happens has a cause, would be just as contingent as the experience itself: its universality and necessity would then be merely feigned, and would have no true
universal validity, since they would not be grounded a priori but only on induction (A195-96 B240-41). Whatever one may think of the argument underlying
the claim of this oft cited passage, it seems clear that Kants target is the view
that the concept of causality has an empirical origin and the principle that everything that happens has a cause is an inductive generalization, which is precisely
what Hume claims in T1.3.3.

33

See my Kants Theory of Taste, Cambridge: Cambridge University Press, 2001,


Chapter 1; Kants Transcendental Idealism (2004), Chapter 15; and especially
Reective Judgement and the Application of Logic to Nature: Kants Deduction
of the Principle of Purposiveness as an Answer to Hume, in Strawson and Kant,
edited by Hans-Johann Glock, Oxford: Clarendon Press, 2003, 169-83.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

541

The problem is that this response, like the arguments of the Analogies as a whole, rests upon a substantive thesis concerning the nature
of human cognition, namely, that such cognition is a discursive activity
consisting in the application of general concepts to sensibly given data.
Consequently, for Kant the cognition of an event requires an interpretative act through which the given sensible data (successive perceptions)
are brought under a rule (the schema of causality), whereby these data
are taken as perceptions of successive states of an object. Thus, just as
Humes challenge to the causal principle rests upon his own theory of
ideas, so Kants reply to Hume relies heavily on what might be termed
his discursivity thesis.
Since this thesis is extraordinarily complex and controversial, I cannot pretend to do justice to it here.34 Instead, I shall limit myself to the
more modest task of clearing up two possible misunderstandings of
Kants view, which bear directly on the point at issue with Hume and
which may have been suggested by the highly schematic rendering of
Kants argument provided above.
The rst of these stems from an ambiguity in the notion of an
objective succession. This can mean a succession either of events or
of the event-stages that constitute an event. If one takes it in the former sense and combines this with the thesis that all objective succession
is governed by the principle of causality, then one seems led to Schopenhauers classical reductio that, on this view, all successive events are
related as cause and eect.35 If one takes it in the latter sense, however,
this untoward consequence does not arise, since the successive stages
constituting an event are not themselves related as cause and eect. All
that the argument, so construed, requires, is that this succession has
some antecedent cause, which may, though certainly need not, involve
a prior state of the same object. Moreover, I think it clear that this is
how objective succession is understood in the argument of the Second Analogy.36
The second and potentially more serious possible source of misunderstanding is the talk of objectivity in terms of interpretation, or

34

35

For my account of the discursivity thesis see Transcendental Idealism, 12-16 and
passim.
See Arthur Schopenhauer, Uber die vierfache Wurzel des Satzes vom zureichenden
Grunde, 23, Samtliche Werke, ed. By J. Frauenstadt, vol.1, Leipzig: FU Brochaus,
1919, 85-92.

36

See, for example, Paul Guyer, Kant and the Claims of Knowledge, Cambridge: Cambridge University Press, 1987, 240.

542

HENRY E. ALLISON

subsumption under a rule.37 The problem with such language is that


it might suggest the attribution to Kant of the manifestly absurd view
that thinking makes it so that is, that the human understanding
somehow has a magical power to generate objective successions and
their causes simply by applying a rule (in this case the schema of causality). Consequently, if Kants reply to Hume and, more generally, the
entire argument of the Transcendental Analytic, is to be taken seriously, it is essential to see that Kant is not committed to any such
view.
To begin with, the basic idea underlying Kants procedure, which
is implicit in rather than argued for in the Second Analogy, is that a
discursive understanding such as ours relates its representations
(intuitions) to an object or, equivalently, makes a claim of objective
validity, by subsuming these representations under a category (more
precisely its schema). As a result of this subsumption, the unication or
synthesis of representations, say the successive perceptions of a ship
in motion, is deemed to hold not merely for a particular consciousness
but for consciousness in general, that is, for any discursive cognizer
presented with the same sensory data. Thus the formula of the Prolegomena: Objective validity and necessary universality (for everyone)
areinterchangeable concepts (4: 298). Otherwise expressed, a claim
of objective validity (in the language of the Prolegomena a judgment
of experience) involves a demand of universal agreement. It states not
merely that this is how things seem to me but how any discursive cognizer ought to conceive the matter. The function of the category or
schema under which the representations are subsumed is to provide a
warrant for this demand. In the case of the Second Analogy, the
object is the alteration of some entity and the schema of causality
may be said to provide the form of the cognition or experience of such
an alteration. Indeed, this is precisely the point of step 5 of Kants
argument as described above.
Although this analysis, with its focus on the a priori conditions of
cognition, obviously diers dramatically from Humes radical empiricism, it is perhaps worth pointing out at this juncture that they share
at least one signicant feature, namely, the recognition of the impossibility of standing, as it were, outside ones representations in order to
37

Although Kant does not typically use terms such as interpretation in these contexts, he does at one point remark that at least part of the task of our power of
cognition [the understanding] consists in spelling out appearances according to a
synthetic unity in order to be able to read them as experience (A314 B370-71).
In Kants Transcendental Idealism, 186-89, I discuss the interpretive role of the
imagination with regard to perception. Needless to say, inasmuch as such interpretation is governed by a priori rules, it has little in common with the familiar Nietzschean conception.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

543

compare them with an objective state of aairs.38 In this respect, they


are both internalists, since for both thinkers what is objective must
be determined on the basis of criteria or conditions that are immanent
to consciousness. The dierences between them turn on the nature of
these criteria or conditions. For Kant they are a priori rules, whereas
for Hume they are provided by associations based on past experience.
More germane to the present point, however, objective validity, as
Kant conceives it, is not equivalent to truth. Kant indicates this in at
least two important places in the Critique. The rst is in the discussion
of the Second Analogy, where he suggests that the Analogies provide
merely formal conditions of empirical truth (A191 B236), which
strongly suggests that he views them as necessary but not sucient conditions of the latter. That is to say, while no empirical judgment can violate the a priori constraints on possible experience imposed by the
Analogies, conformity to these constraints is not sucient to determine
the veridicality of a judgment. The latter is an empirical, not a transcendental matter, and as such is determined by the usual empirical means.
In the case of causal judgments governed (but not determined) by the
schema of causality, these means might very well include Humes Rules
by which to judge of causes and eects (T 1.3.15; SBN 173-76), which
for Kant would presuppose the causal principle of the Second Analogy.
The second relevant passage is from 19 of the B-Deduction, where
Kant characterizes a judgment as a relation [of representations] that is
objectively valid, as contrasted with a relation of the same representations according to the laws of association to which he attributes merely
subjective validity (B142). Clearly, in making objective validity into a
denitional feature of judgment, Kant is not suggesting that every judgment, simply qua judgment, is true. His point is rather that every judgment makes a claim to truth or has a truth value, whereas a union of
representations produced by association can be neither true nor false.
The same applies, mutatis mutandis, to the subsumption of representations under the schema of causality, which is, after all, itself just a
judgment of objective succession.
Having disposed of these misconceptions, which threaten to block
any serious consideration of Kants views regarding causality, we are
38

In the Analogies, which are concerned with objective time-determination, Kant


tends to express this point by noting that time itself cannot be perceived. I discuss this at several junctures in Transcendental Idealism (See pp. 216-17, 236-41,
264-66). Strangely enough, Kant does not appeal to the unperceivability of time in
the main argument for Second Analogy, though he does so in the arguments for
the other two, as well as in the argument for the general principle of the Analogies.
Nevertheless, I believe that it plays a central, albeit somewhat subterranean role
there as well. Moreover, it has an analogue in Humes denial that we have a distinct impression of time (to be discussed below).

544

HENRY E. ALLISON

nally in a position to return to our central question: does Kant provide an effective response in the Second Analogy to Humes denial of
a priori status to the causal principle? Although the discussion cannot
be divorced completely from the fore-mentioned global issues regarding
cognition, I believe that the major point at issue may be put fairly succinctly: can events, as dened above, be directly perceived or just
seen, or does the judgment that something has happened require some
kind of interpretive act? Kant, as we have seen, arms the latter,
whereas Hume is committed by both his account of impressions and
his theory of philosophical relations to the former.
According to Humes account of impressions, the perception of an
event must be a complex impression, such as that of a billiard ball in
motion after being struck by another ball or a die lying on one of its
sides after being tossed. According to his theory of philosophical relations, those of time and place (together with identity) are matters of
perception rather than reasoning, which, as such, require nothing
beyond a mere passive admission of the impressions through the
organs of sensation (T1.3.2.2; SBN 73). Moreover, this thesis about
event-perception seems warranted, if we understand it in the Humean
manner. Since such perception involves nothing more than having a
complex impression, and since, ex hypothesi, the mind is passive with
respect to its impressions (both simple and complex), events may be
said to be just seen, rather than inferred or judged. An act of mind
is only required subsequently, if we attempt to infer the existence of
another event as either its cause or eect.39
Nevertheless, I think it doubtful that the perception of an event can be
adequately characterized in this manner. What it leaves out is the dynamical element in such perception, which is emphasized by Kant. On this
model, to perceive an event or, more precisely, to cognize one, is not simply to perceive a state of affairs that happens to succeed an earlier one,
but to perceive it as succeeding the previous state. And this does seem to
require the interpretation (or taking) of a sequence of perceptions (which
is all that is strictly given) as the perception of a succession in an object.
Moreover, whatever the obstacles to reconciling it with his ofcial
theory, it is difcult to see how Hume could avoid acknowledging the
need for something like an interpretation in this sense, at least not if he
39

Apart from reference to Hume, the objection that we can just see events and, therefore, have no need to appeal to any transcendental principles is sometimes raised
against Kant. See, for example, H.A. Prichard, Kants Theory of Knowledge, reprint, New York and London: Garland Publishing Company, 1972, 294-6, and Jeffrie Murphy, Kants Second Analogy as an Answer to Hume, Ratio 11 (1969),
75-78. My own treatment of this issue has been inuenced by L.W. Becks response
to Murphy. See especially Becks On Just Seeing the Ship Move, Essays on
Kant and Hume, 136-40.

WHATEVER BEGINS TO EXIST MUST HAVE A CAUSE OF ITS EXISTENCE

545

wishes to explain how, on the basis of observed regularities, we assign


causes, form beliefs regarding particular causal connections, and
acquire the custom of expecting similar event-sequences in the future.
For, as I have already argued, what normally requires a cause is not
the being of something in a certain state but its getting to be in that
state, that is, its alteration or, in Humean terms, its acquisition of a
new modication of existence. But then Hume must confront the problem of accounting for the possibility of distinguishing between having
successive perceptions of a static object and having the perception of
either successive states of an object or successive events, which is not to
be confused with the problem posed by Anscombe of distinguishing
between something coming into existence at a certain place and migrating there from elsewhere.
Finally, although I cannot develop the point here, the difculty seems
to be exacerbated by Humes account of time. In his most illuminating
treatment of the topic, Hume appeals to the example of ve successive
impressions of notes played on a ute. His point is that the idea of time
is not derived from an additional distinct impression (since there is
none), but arises entirely from the mind noticing the manner, in which
the different sounds make their appearance (T1.2.3.10; SBN 37). Since,
as Hume points out, this manner of appearing is as successive, this commits him to the view that the idea of time arises from the perception of
a succession. But herein lies the problem; for in addition to the apparent circularity (the awareness of a temporal succession presupposes the
idea of time), it seems that Hume simply helps himself to a perception
of succession. In other words, the having of ve successive note-perceptions is one thing and the perception of a succession of ve notes quite
another. And if this is correct, it follows that, even apart from the question of event-perception, there is need for an interpretive act as a condition of the possibility of the awareness of a temporal succession.40 In
short, the Humean mind enters the story one step too late.41
40

This act is what Kant described in the A-Deduction as a three-fold synthesis (or at
least the rst two parts thereof). The basic point is that in order to take them as
constituting a successive series of notes, the mind must unite its successive perceptions, which, in turn, requires that it retain or reproduce its awareness of its past
perceptions of the notes as it successively attends to the later ones. For, as Kant
puts it, [I]f I were always to lose the preceding representations...from my thoughts
and not reproduce them when I proceed to the following ones, then no whole representation...could ever arise (A102).

41

I wish to thank an anonymous referee for this journal, Angela Coventry, who was
my commentator at a Hume session at the 2006 Pacic Division meeting of the
APA, as well as members of the audience at that session and at subsequent colloquia at UCI and UCSD for their helpful comments regarding various versions of
this paper. I have attempted to incorporate many of their suggestions and calls for
clarication into this nal version.

546

HENRY E. ALLISON

Das könnte Ihnen auch gefallen