Kant on Euclid: Geometry in Perspective
Prof. Stephen Palmquist,
D.Phil. (Oxon)
Department of Religion and Philosophy
Hong Kong Baptist University
I. The Perspectival
Aim of the first Critique
There is a common assumption among philosophers, shared even by many
Kant-scholars, that Kant had a naive faith in the absolute validity of
Euclidean geometry, Aristotelian logic, and Newtonian physics, and that his
primary goal in the Critique of Pure Reason was to provide a rational
foundation upon which these classical scientific theories could be based.
This, it might be thought, is the essence of his attempt to solve the problem
which, as he says in a footnote to the second edition Preface, "still
remains a scandal to philosophy and to human reason in general"--namely,
"that the existence of things outside us...must be accepted merely on faith,
and that if anyone thinks good to doubt their existence, we are unable to
counter his doubts by any satisfactory proof" [K2:xxxix].1 This assumption, in
turn, is frequently used to deny the validity of some or all of Kant's
philosophical project--or at least its relevance to modern philosophical
understandings of scientific knowledge. Swinburne,
for instance, asserts that an acceptance of the views expressed in Kant's first
Critique "would rule out in advance most of the great achievements
of science since his day."2 Such
assertions--and in particular the claim that Kant's argument for the
transcendental ideality of space in the "Transcendental Aesthetic" of
the first Critique can be ignored because it rests on an undefended
acceptance of Euclidean geometry --are such a commonplace that there is no need
to quote further examples.
But is the "satisfactory proof" Kant offers in the first Critique
inextricably tied to the necessary validity of Euclidean geometry, Aristotelian
logic, and Newtonian physics, as is so often assumed? A deeper
understanding of both the structure and the purpose of Kant's
argument, as well as a proper recognition of the context in which he
employs such traditional theories, may reveal quite a different picture.
In this article I will begin the task of setting right a travesty, on the basis
of which Kant's first Critique has so often been rejected prematurely,
by carefully examining the function of Kant's treatment of Euclidean
geometry in his theoretical system. In short, I will argue that, far from
assuming a kind of absolute validity for the classical theories of Euclid,
Aristotle, and Newton, Kant ties their views to a well-defined perspective in
such a way that their validity is actually limited, and that in so doing
he actually prepared the way (sometimes with surprising foresight) for
modern developments in geometry, logic, and physics.
Throughout this article I will utilize a method of interpreting Kant's Critical
System which I have developed in detail elsewhere [see note 3]. The key
to this method is the recognition of the ubiquitous influence of the
"principle of perspective" in Kant's thought, as it is expressed both
in the overall structure of his System and in the myriad of distinctions within
that System. This principle requires the philosopher to consider the
general presuppositions implied by the context of any given question or
problem, together with the rules that guide one to a solution, before
attempting to deal with it. It implies that different, though equally
valid, answers can be given to the same question if different
"perspectives" are assumed. (Kant himself never explicitly
states this as a principle, nor does he use the term "perspective" in
any technical sense. However, he does make frequent use of a number of
terms, such as "standpoint", "point of view", "way of
thinking", etc., in precisely the way indicated by my use of the term
"perspective".) Kant's three "general perspectives",
or standpoints, are the theoretical, practical, and judicial, and each
is developed, respectively, in one of his three Critiques. But the
subject-matter elaborated from any one of these standpoints can itself be
considered from several different angles, depending on the assumptions made at
a given stage in its development. Accordingly, within each
standpoint--and in this article I will be dealing only with the theoretical
standpoint--Kant develops four subordinate perspectives: the
transcendental, logical, empirical, and hypothetical. The
"Transcendental Doctrine of Elements" in the first Critique
has four main subdivisions, which correspond, respectively, to these four
perspectives: the "Aesthetic", the "Analytic of
Concepts", the "Analytic of Principles", and the
"Dialectic".3 With this perspectival framework in mind, let us now turn to the
question at hand and attempt to discern Kant's actual intentions vis-a-vis the above-mentioned classical doctrines, paying
special attention to his view of Euclidean geometry.
II. The Transcendental Character of Geometry
In the second edition Preface to the first Critique, Kant says his
"Copernican revolution" proceeds "in accordance with the example
set by the geometers and physicists" [K2:xxii]. Their example
(together with the logicians [viii-ix] is appealing because in Kant's opinion
logic, mathematics and physics have all "entered upon the secure path of a
science" [vii], precisely because they "have to determine their
objects a priori" [x]--an insight which may seem strange to us,
since "science" is now normally considered to be primarily empirical,
but which is crucial in understanding Kant's intentions. Kant clearly
regards the established (Euclidean, Aristotelian and Newtonian) models as
having attained a kind of absolute certainty, but this does not mean, as
is generally assumed, that he therefore rejects the possibility of other,
equally valid models being developed.4 On the contrary,
as we shall see, Kant's contribution was to show that these classical models
can be regarded as absolutely valid only when certain limitations are
placed on them, by viewing them from a particular perspective.
Indeed, two sentences after referring to his use of geometers and physicists as
exemplars, he adds that "pure speculative reason has this peculiarity,
that it can measure its powers according to the different ways in which it
chooses the objects of its thinking [i.e. its different perspectives on
objects]" [K2:xxiii, emphasis added]. And this should serve at least
as a hint that Kant intends his System to provide a philosophical basis for
regarding geometry, logic, and physics as capable of different, but equally
valid, non-transcendental formulations.
An even clearer hint comes in Kant's first published writing (1747), when he
poses the problem: "The ground of the threefold dimension of space
is still unknown" [K1:23]. He admits that the law which determines
that space "has the property of threefold dimension...is arbitrary, and
that God could have chosen another... law [from which] an extension with other
properties and dimensions would have arisen. A science of all these
possible kinds of space would undoubtedly be the highest enterprise which a
finite understanding could undertake in the field of geometry"
[24]. He suggests the law in question might be "that the strength
of the action [of substances] holds inversely as the square of the
distances" [24]. However, he clearly recognizes that "the
necessity of the threefold dimension" is not a logical necessity,
"but rather...a certain other necessity which I am not as yet in a
position to explain" [23]. His suggestions in this initial inquiry,
he says, "may serve as an outline of an inquiry which I have in prospect"
[25]. Thirty-four years later this "prospect" finally became a
reality!
The necessity of the three-dimensional nature of space is explained in K2 as
resulting from the fact that space is the pure form of our sensible intuition
[K2:37-45]. Martin gives a detailed account of the mathematical
implications of this theory. He explains that the intuitive character of
mathematics means, for Kant, "that mathematics is limited to objects which
can be constructed" [M1:23]. In other words, Kant's mature position
is that intuition "limits the broader region of logical existence...to the
narrower region of mathematical existence" [25]. Nevertheless, Kant
did not give up his early insight regarding the logical possibility of
non-Euclidean geometries: "there can be no doubt that it was clear
to Kant that in geometry the field of what is logically possible extends far
beyond that of Euclidean geometry" [23]. Indeed, he claims that
"under the Kantian presuppositions it is not only possible but necessary
to assume the existence of non-Euclidean geometries" [18]. Martin adds:
"Non-Euclidean geometries are logically possible but they cannot be
constructed; hence they have no [real] mathematical existence for Kant and are
mere figments of thought."5
What Martin and other commentators fail to recognize is that, just because they
are "figments of thought" does not mean that a Kantian must
immediately reject non-Euclidean geometries as useless speculation. His
main point is not that they are physically impossible, but that it is
impossible to form an image or picture of what they would look
like. Thus, after arguing in K1 that multi-dimensional physical
spaces are logically possible, Kant adds the important qualification
that "in anything representable through the
imagination in spatial terms, the fourth [spatial dimension] is an
impossibility" [K1:23]. He then connects them directly to the Leibnizian conception of many possible "worlds",
which, he argues, can be regarded as not just possible, but
"probable", only on the assumption "that those many
kinds of space, of which I have just spoken, are likewise
possible"[K1:25]. Kant is not here making a positive claim about the
real existence of non-Euclidean geometries, but only suggesting that we keep an
open mind (which, on the traditional interpretation, he himself did not
do). Before we investigate in more detail this issue of the conceptual
status of non-Euclidean geometries, let us examine the actual importance of
Euclidean geometry to the overall validity of Kant's System.
In the Aesthetic of K2 Kant seems to argue at several points from the
validity of Euclidean geometry to the transcendental ideality of
space. He says in K2:41, for example, that "geometrical properties
are one and all apodeictic, that is, they are bound
up with the consciousness of their necessity; for instance, that space has only
three dimensions."6 Thus, he argues,
"the only explanation that makes intelligible the possibility of
geometry" is that space is a formal condition of experience, imposed on
objects by the human mind. Later, he hinges his argument again on a
similar reference to the need to explain the apodeictic
certainty of mathematical propositions, a type of certainty which "is not
to be found in the a posteriori" [K2:57]. And in K2:64-66 he
provides his most lengthy argument of this sort: he takes the certainty
of geometrical propositions as the starting-point (presupposed on the basis of
his arguments in K2:14-17), and argues that this requires us to regard space as
a "pure a priori intuition". Although Euclid is never
mentioned by name in the entire Critique, such passages clearly indicate
that Kant believed Euclidean geometry had attained a kind of certainty and
necessity which places it beyond question.
As a result, Strawson argues that in the Aesthetic
"the doctrine of the transcendental subjectivity of space rests on no
other discernible support that [sic] that provided by the argument from
geometry"!7 Kant himself,
however, does not present the few brief arguments "from geometry" as
proofs of his theory of space, but rather he presents his theory of
space as a support for a proper explanation of the necessity of
geometrical propositions. The difference really rests on a different
conception of the importance of arguments in the philosophical task
itself: analytic philosophers such as Strawson
regard a good theory as one that is based on good arguments; synthetic
philosophers such as Kant, by contrast, regard a good argument as one which arises
out of a good theory. In other words, Kant's arguments "from
geometry" are not intended to serve as the basis for anything;
rather, they are intended to demonstrate that a "Copernican" view of
space as transcendentally ideal--i.e., as a subjective, a priori form of
experience--provides the basis for an explanation of the otherwise inexplicable
necessity of geometrical propositions.
Moreover, Kant does not take an entirely uncritical attitude towards Euclidean
geometry. He uses it in the above-mentioned ways not because he has no
other way to defend his theory of the transcendental ideality of space, but
because the consensus of opinion in his day was that Euclidean geometry was undeniably
true; hence it could be taken as an unproblematic premise in a way which would
be impossible today. However, as we shall see, that very change is due in
large part to Kant. For even though he takes the certainty of geometrical
propositions as the starting point for a few of his arguments, his conclusion denies
the validity of the traditional belief that such propositions apply to a
physical reality called "absolute space" [K2:54-56]. Therefore,
rather than siding with the consensus of opinion on the relevance (or irrelevance)
of Kant's theory of geometry for contemporary philosophy of science, we must
now look more closely at the perspective from which he regards geometrical
propositions to be necessarily true.
Is Kant assuming in the Aesthetic that Euclidean geometry provides the
true explanation of the real structure of physical space? Let us look
more closely at Kant's actual position. Kant concludes his "Transcendental
Exposition of the Concept of Space" with a summary, which is intended to
serve as a warning:
The transcendental
concept of appearances in space...is a critical reminder that nothing intuited
in space is a thing in itself, that space is not a form inhering in things in
themselves..., and that what we call outer objects are nothing but mere
representations of our sensibility, the form of which is space. The true
correlate of sensibility, the thing in itself, is not known, and cannot be
known, through these representations; and in experience no question is ever
asked in regard to it. [K2:45, emphasis added]
Here Kant is clearly warning the reader not
to regard the arguments of the Aesthetic, which adopt the transcendental
perspective, as applying also to the empirical perspective. For,
as he puts it quite bluntly, the sorts of questions he asks in this part of the
Critique would not even arise if we limited our attention to the
empirical perspective. Yet this warning has been overlooked or ignored by
most of Kant's critics, with the result that Kant's position in the Aesthetic
is probably the most frequently rejected part of the entire Critical System.
One of the most unfortunate results of the tendency to ignore Kant's warning
against neglecting the perspectival character of his
arguments is that he is interpreted as saying that Euclidean geometry is
necessarily true of the physical world. In fact, a careful reading of the
Aesthetic reveals that he never says anything of the kind! Rather, his
whole argument is intended to draw the reader away from such empirical
questions and towards questions concerning what is "bound up with [human]
consciousness", and is therefore "apodeictic"
[K2:41] in a completely non-physical (or meta-physical) way. The
Aesthetic can only be understood as presenting a coherent argument once we
recognize that in it Kant is not doing physics! Rather, he expects us to
join with him in limiting our attention to the transcendental perspective.
Viewing it in this way enables us to see that Kant is using geometry as an
example--a test case--and not as an essential element in his system. Kant
himself makes this quite clear when he introduces his argument from geometry in
K2:64-66 in the following way: "To make this certainty [i.e. the
certainty of his view that space is an a priori form of intuition] completely
convincing, we shall select a case by which the validity of the position
adopted will be rendered obvious..." [K2:64-65]. Obviously, he does
not think the transcendental ideality of space depends at all on the
validity of Euclidean geometry, but rather, vice versa!
In what sense, then, does Kant believe Euclidean geometry can rightfully claim apodeictic certainty? Certainly not, as is
often assumed, by pointing to the empirical world and saying "See, it's
true!" That would be to remain in the empirical perspective,
which, as he says in K2:A24, can only tell us about "the contingent
character of perception". Rather, its necessity can be explained
only from the transcendental perspective: only by regarding
geometrical propositions as describing the way in which we must present
space to ourselves in our sensible experience. Kant is arguing that
Euclidean geometry describes the form of our perception of things in
space, not the way they are actually related. And this, as we
shall see, is not only plausible even today, but it leaves open a place
for other geometries which might adapt the Euclidean model in
such a way that it can apply to empirical reality itself (i.e. to the physics
of space).
Just what, then, is Kant's attitude towards the empirical applicability of
Euclidean geometry, and how important is the Euclidean structure of physical
space to the validity of Kant's theory? Moreover, what exactly does Kant
mean by his doctrines of "outer sense" and "pure
intuition"? Kant begins his explanation in K2:37 by saying: "By
means of outer sense, a property of our mind, we represent to ourselves objects
as outside us, and all without exception in space." This
"metaphysical exposition" is not intended to describe the way we experience
space or objects in space; that would be an empirical concern. Rather, it
is intended to describe the nature of the human subject. Without straying
into the difficult subject of the meaning of "pure intuition",8 we can point out that
one of the main points Kant is making in the Aesthetic is that, viewed from the
transcendental perspective, space and time are two distinct forms of
sensibility inhering in the human subject, despite the fact that in experience
we always find them together. (The significance of this point will be
discussed in section III.) The real thrust of Kant's argument is simply
to point out that transcendental reflection requires us to distinguish between
these two sources of material for knowledge, and to recognize that each, viewed
separately, has a fixed, predetermined form. But this form is
transcendental and relates to appearances; it is valid, therefore,
"only in us" [K2:59], and leaves open the question of whether or not
exactly the same forms will hold true when objects are regarded from the
empirical perspective as being "outside us" [as in K2:37; see also
275-276].
According to Kant, therefore, Euclid's geometrical system is a transcendental
abstraction from actual experience, and only because of this fact--not
because the world is "really" structured in this way--can his system
claim certainty and necessity. Kant is claiming that if we abstract space
and time separately from our experience of the real (empirical) world, and
consider the necessary requirements for perception, then the resulting picture
of this abstract, spatial world (as opposed to the empirical world of
space-time) will be Euclidean. Kant regards this as a kind of "brute
fact" about how the human mind is structured [see e.g. K2:42], in
much the same way as he views the categories [K2:145-146,150-151], the schemata
[K2:180-181], and the moral law [see e.g. K3:31] as brute facts; but we have no
way of knowing whether or not they hold true for all rational beings
[see e.g. K2:71-72; K3:72]. This conclusion is considerably less
problematic than the view traditionally attributed to Kant, that physical space
is actually Euclidean, because most geometers and physicists would readily
agree that the only world we can "picture" with our imagination, and
so also the ordinary world as we see it, is Euclidean.9
If, then, we assume that Kant was correct in his assumption that our
sensibility is limited to a Euclidean picture of the world, what are the
implications of this theory for geometers themselves? Kant's view is that
the Euclidean nature of space is actually a transcendental condition for the
very possibility of our perception of space, so that spatial objects viewed
as appearances must assume this form. But Kant clearly understands
that this transcendental perspective is primarily of interest to the
philosopher; if Kant had ever had the opportunity to share his lunch with an
Einstein, he would have readily admitted that the scientist is fully justified
in viewing the world empirically, so that physical objects can be treated
as independent "things in themselves".10 This implies that
the geometer (or physicist) who accepts Kant's transcendental perspective does
not need to assume that the physical world itself conforms to the way we
must view it. That is, Kant's theory, by clearly distinguishing between
the transcendental and empirical perspectives, and by associating Euclidean
geometry exclusively with the former, actually raises the question as to
whether or not some alternative geometry, though not picturable
to our sensibility, might conform more closely to the way objects are actually
structured in (empirically) real space.
The chief objection against this interpretation of Kant's intentions is, no
doubt, that it flies in the face of the almost universally accepted assumption
that Kant is trying to guarantee the applicability of Euclidean
geometry to the physical world. Unfortunately, Kant himself never clearly
states just what his intentions are in this respect. As a result, the way
of reading Kant which was natural in the eighteenth century, when the question
of the actual structure of the physical world would hardly be raised by most
readers, because Euclidean geometry was generally accepted as unquestionably
true, has become the traditional interpretation of Kant's own intentions, and
has never been significantly challenged. Thus, for example, when Kant
says in K2:121 that "the concepts of space and time", regarded as
pure intuitions, "independently of all experience...make possible a synthetic
knowledge of objects", the natural assumption is that "objects"
here refers to real empirical objects. Yet this is not Kant's
intention at all: the "objects" of which we have synthetic a
priori knowledge by means of Euclidean geometry are exclusively appearances,
and as such are viewed transcendentally.11
Strawson's analysis of Kant's theory of geometry
provides a good example of the traditional interpretation. He explains that
geometry can be viewed as relating either to a set of phenomenal (and thus unfalsifiable), pictures, or to the logic behind them, or
to their application to objects in physical space. He states "that
Kant's theory of pure intuition can be construed as a reasonable account of the
nature of geometry in its phenomenal interpretation" [S1:284]. He
then asserts that Kant intended "to use his insight into the necessities
of phenomenal geometry to resolve... the difficulty created by the
apparently necessary application of Euclidean geometry to physical space"
[284-285]. Thus, he continues,
Kant's fundamental
error...lay in not distinguishing between Euclidean geometry in its phenomenal
interpretation and Euclidean geometry in its physical interpretations...
Because he did not make this distinction, he supposed that the necessity which
truly belongs to Euclidean geometry in its phenomenal interpretation also
belongs to it in its physical interpretation. He thought that the
geometry of physical space had to be identical with the geometry of
phenomenal space. [285]
Significantly, Strawson does not refer to a single text
to support his acceptance of this traditional interpretation--i.e. his
assumption that Kant's notion of pure intuition corresponds to phenomenal
geometry and that its empirical application entails an identification of
physical geometry with phenomenal geometry. By contrast, if we regard
transcendental geometry in the way I have suggested, then Kant's doctrine of
pure intuition "accounts" for phenomenal geometry only in the sense
that it serves as its transcendental foundation. That is, the empirical
application of pure intuition gives rise to phenomenal geometry, but is
never directly related by Kant to physical space. Thus, Strawson's "fundamental error" lay in not
distinguishing between Euclidean geometry as transcendental and Euclidean
geometry as phenomenally applied. Kant's silence on the subject of
physical geometry need not be interpreted as an identification of phenomenal
with physical, but may simply reflect his recognition that physical geometry
(as a posteriori) is a subject which need not be addressed by the
transcendental philosopher.
Strawson does quote K2:155n to make quite a different
point about the construction of phenomenal figures [S1:289]. Yet this
text actually denies the very (traditional) assumption which Strawson takes for granted! In K2:155n Kant says:
Motion of an object in space
does not belong to a pure science, and consequently not to geometry. For
the fact that something is movable cannot be known a priori, but only
through experience. Motion, however, considered as the describing of a
space, is a pure act of the successive synthesis of the manifold in outer
intuition in general by means of the productive imagination, and belongs not
only to geometry, but even to transcendental philosophy.
Kant's main point (which Strawson
ignores) is that "pure [i.e. a priori] science" must be carefully
distinguished from applied, or empirical science, so that once
"facts" such as some particular motion are taken into consideration,
we are no longer even talking about geometry (whether Euclidean or
non-Euclidean).12 In other words,
Kant seems to be saying in this footnote that the applicability of
geometry to objects in physical space is a problem which lies outside the
bounds of geometry (and transcendental philosophy)! And this
implies, of course, that Euclidean (or any other pure) geometry does not necessarily
hold true of physical objects when they are viewed as objects of empirical
science.
This interpretation of K2:155n makes sense only if we remember that Kant
defines geometry as "a science which determines the properties of space
synthetically, and yet a priori" [K2:40; see also 120]. Given
this definition, of course, Kant could not admit any non-Euclidean geometries,
because both of the types of geometry which Russell describes (viz. logical and
physical [see note 11]) would not count as "geometry" in Kant's
special sense. Only Euclidean geometry is synthetic and a priori, so only
it is relevant to the transcendental philosopher. When this is
understood, the question of whether or not physical space is Euclidean can be
seen in its proper perspective, as a side issue relating not to the validity of
transcendental philosophy, but only to the question of the significance of Euclidean
geometry for empirical science.
III. The Empirical Application of Transcendental
Geometry
Does this mean that the traditional interpretation of Kant, according to which
he is attempting to give a transcendental guarantee that Euclidean geometry
must be true of the physical world, is a groundless myth? Certainly not,
for there are many passages in which Kant is indeed saying something along
these lines. A prime example comes in the section entitled
(significantly) "Axioms of Intuition", where Kant says:
This transcendental
principle of the mathematics of appearances [i.e. that "All intuitions are
extensive magnitudes" [K2:202]] greatly enlarges our a priori
knowledge. For it alone can make pure mathematics...applicable to objects
of experience. Without this principle, such application would not be thus
self-evident; and there has indeed been much confusion of thought in regard to
it. Appearances are not things in themselves. Empirical intuition
is possible only by means of the pure intuition of space and of time.
What geometry asserts of pure intuition is therefore undeniably valid of
empirical intuition. The idle objections, that objects of the senses may
not conform to such rules of construction in space as that of the infinite
divisibility of lines or angles, must be given up.... [206]
This text seems to lend unequivocal support to the
traditional interpretation: here Kant clearly states his conviction that
the transcendental validity of Euclidean geometry implies a corresponding
empirical validity.
The question is, just what does Kant mean here by "undeniably valid of
empirical intuition"? Is there any way to accept Kant's view
of the determining influence of the transcendental on the empirical, and yet to
preserve some degree of autonomy for the empirical scientist (an autonomy
which in other passages Kant seems to uphold)? Such questions did not
occur to Kant, because the science of his day had not yet fully claimed its
autonomy (from philosophy) as a thoroughly empirical discipline; hence Euclid's
"transcendental" geometry was still believed by nearly all scientists
and philosophers to be the last word on empirical matters as well. Kant
was clearly influenced by his tradition in a way for which he can hardly be
blamed. The important point, however, is that the perspectival
character of his System points us directly to a way of supplementing his own
explicit views in order to account for the modern developments. In this
section I will attempt to demonstrate just how this can be done.
Kant in all likelihood believed Euclidean geometry to give a true account of
the structure of the physical world. Nevertheless, even if we would now
prefer a non-Euclidean account, the essential structure of his theoretical
system can remain in tact, once we realize that the form of our sensible
perception of objects in space, when all time-considerations are
abstracted, is indeed necessarily Euclidean. And in this phenomenal
sense, as a description of how we perceive the world, Euclid's system is
regarded as correct even today. How, then, can a physical theory of
curved space, with its corresponding non-Euclidean geometry, be compatible with
the phenomenal validity of Euclidean geometry? Kant provides us with a
way of solving this difficult problem.
When Kant says in K2:206 that the geometry of pure intuition is
"undeniably valid" for empirical intuition as well, he is not denying
that the physicist can conceive of a non-Euclidean geometry which
actually holds true for aspects of space which human beings cannot perceive.
On the contrary, the objects of "empirical intuition", which he is
claiming must necessarily conform to Euclidean geometry, are objects of
"possible experience", by which Kant means "objects which are
within the range of human perception". And the fact which is too
often neglected in most accounts of Kant's supposedly miserable failure to
foretell the future by providing for modern scientific advances is that these
advances all have to do with "viewing" physical objects in their
extreme manifestations--i.e. quantities which are either too small for
human beings to perceive (as, e.g., in quantum physics) or too large for
human beings to perceive (as, e.g., in the application of relativity theory to
astrophysics). Yet, giving proper emphasis to this fact enables us to
locate the real problem in Kant's view, which is simply that he failed to
acknowledge that the objects of empirical science are not limited to perceivable
objects. In other words, Kant's analysis of the limits of human
knowledge neglected to consider the status of empirical sciences which
experiment with quantities too small or too large ever to be intuited by
the human sensibility. For example, the modern theory of physical space
as "curved" is based on conceptions and calculations, not
on perceptions of a space which actually appears curved. By
simply recognizing that physical science does not always require direct
intuition of the objects on which it experiments, we can therefore reconcile
the modern views concerning the geometry of physical space with Kant's view of
the transcendental necessity of our perception of Euclidean
space.
Moreover, in the geometry of curved space, the perspective-lessness of the observer (or the unobservability
of the perspective) is of utmost importance. For in order to perceive
our space as curved, we would have to be able to observe it from the
standpoint of some other space (outside of our own space) which is not
curved. And this is precisely Kant's point about what man cannot
do: his doctrine of pure intuition is intended to drive home the fact
that it is "solely from the human standpoint that we can speak of
space" [K2:42]. The equivalent requirement in Euclidean geometry
can be accomplished, however, by adopting, as it were, the standpoint of time
and looking at space from this perspective as nontemporal.13 This, in Kantian
terms, is the crucial difference between Euclidean and non-Euclidean
conceptions of space: there is no human standpoint from which the latter
can be perceived.
Thus, as we have already seen, Euclidean geometry is a (transcendental)
abstraction from experience: it arises only when the subject, his motion
and his time, are removed from observation (and in this sense, adopted as the
standpoint). "Abstracting time" from our consideration of
experience really means, therefore, regarding space as an object of spatial appearance--an
intuited "picture" of the real world. In other words, Kant's
theory is telling us that the human perceiver breaks experience up into
temporal and spatial components: the reality we experience exists as a
unified, space-time continuum; but the perceiver determines it in such a
way that it appears to be Euclidean. Thus, the empirical object
viewed from the transcendental perspective as in us (i.e. as an
appearance) is Euclidean, even though that same object viewed from the
empirical perspective as outside us (i.e. as a phenomenon in space-time) might
be non-Euclidean.
When the scientist intentionally refuses to submit to the natural human
tendency to perceive time and space as separate (i.e. refuses to let the
inner-outer distinction influence his science), then the problem of
determining the structure of space is immediately transformed from a static,
transcendental inquiry, into a dynamic, empirical inquiry. (As a result,
space is, in a sense, dehumanized.) Any resulting non-Euclidean theory of
space is bound to be "counter-intuitive" (i.e. not picturable) for precisely the reasons Kant gives in his defence of the transcendental character of Euclidean
space: since Euclidean geometry is, as Kant argued, the form of our
intuition of space, other geometries can never be perceived, but only conceived,
even if we subsequently discover that a non-Euclidean explanation, though
purely conceptual, nevertheless enables us to explain certain phenomena better
than a Euclidean one. This is because the other theories can be drawn
out from experience, but are not read into experience by the human
subject. For this very reason (i.e. by definition) only Euclidean geometry can
be both synthetic (true of our abstract, nontemporal
representation of our experience) and a priori (deriving its truth from a
source other than experience); the synthetic applicability of all other
geometries can be known only a posteriori (after collecting data based on
cumulative experiments). In other words, our knowledge of the applicability
of any non-Euclidean geometry will be contingent because it presupposes
experience, whereas that of Euclidean geometry is necessary because it prefigures
experience. And this is why Kant claims that no non-Euclidean geometry
can exist (i.e. can have real possibility): because it
would have to be a posteriori, and therefore could not be pure (so it
would not be a true "geometry" according to his definition).
What is almost always ignored in discussions of Kant's theory of geometry is
that Kant himself makes some (very "modern") suggestions in the
"Transcendental Dialectic" of K2 concerning the usefulness of
just such non-intuitive theories for science. A theory which is
conceptual but which contains no intuitable content
cannot, he says, be regarded as establishing "empirical knowledge",
but viewed as an "idea" it can serve as an hypothesis with definite
heuristic value. He defines an "idea" as "a concept formed
from notions and transcending the possibility of experience"
[K2:377]. The "transcendental ideas", he explains,
"determine according to principles how understanding is to be employed in
dealing with experience in its totality" [378]. Kant's main effort
in the Dialectic is, of course, to expose the fallacies which arise whenever
such ideas are thought to constitute empirical knowledge. As a result,
his lengthy appendix on "The Regulative Employment of the Ideas of Pure
Reason" [670-698] is often ignored. In it he explains that ideas of
reason can have a legitimate "immanent" use if we treat them "as
if" they are true: "The hypothetical employment of reason is
regulative only [not constitutive of knowledge]; its sole aim is, so far as may
be possible, to bring unity into the body of our detailed knowledge"
[675]. If Kant were alive today, it seems likely that he would apply this
theory of ideas to any attempt to describe physical space as
non-Euclidean. In other words, he would say that if a non-Euclidean concept
of space enables scientists "to bring unity into the body of [their]
detailed knowledge", then they are more than welcome (indeed, encouraged)
to make such a conjecture, provided they never claim to have established empirical
knowledge of its certain truth. And this seems to be quite an
accurate account of the way modern scientists do view such theories!
Returning now to the quote from K2:206, we can see that Kant is not, in fact,
contradicting the perspectival interpretation which I
have been defending. For his claim is that pure intuition determines the
nature of empirical intuition; any consideration of empirical intuition
assumes the transcendental perspective (that of the Aesthetic, where space and
time are separately abstracted from experience), not the empirical perspective
(that of the Analytic of Principles, where space and time are put back into
experience, and where intuitions and concepts are synthesized by the
imagination). And, as we have seen, Kant does not limit scientists to the
empirical perspective, but encourages them to adopt the hypothetical
perspective (that of the Dialectic, where concepts without intuitions are
viewed as regulative ideas).14 So scientists who
do not purport to be examining "objects of the senses" are not bound
by Kant's theory to force the results of their inquiries into a Euclidean
(sensible) mold.
IV. Concluding Remarks on
Kant's Copernican Perspective
The novelty of this conclusion may be surprising at first; yet upon reflection
it should be viewed as a natural implication of Kant's whole Copernican
revolution. For his assumption that the subject determines
(transcendentally) the perceived character of empirical objects always acts as
a two-edged sword: he uses one edge to cut the traditional positions off
from the domain to which they were formerly believed (erroneously) to apply;
and with the other edge he protects those same positions from further
attack by putting them in their proper place. Kant makes his dual motivations
quite clear with respect to metaphysics, morality and religion in
K2:xxx-xxxi. And there is no reason to suppose the implications of his
Copernican perspective on geometry to be an exception. In fact, this
Copernican perspective can be seen working in a remarkably similar way by
comparing Kant's attitude towards Euclidean geometry with his attitude towards
Aristotelian logic and Newtonian physics.15 However, such
comparisons are beyond the scope of this paper.
Even in his earliest essay Kant shows an awareness of the perspectival
distinction we have been examining: "body as mathematically
conceived is a thing quite distinct from body as it exists in nature; and
statements can be true of the former which cannot be extended to the
latter" [K1:140]. And as we have seen, Kant's mature philosophy also
leaves open a place for admitting a valid scope of application for
non-Euclidean geometries, systems which do not impose predetermined forms onto
perceived objects, but which ignore the empirically perceived forms and
view the objects "as if" they participate in a single, space-time
reality. This may be precisely what Kant had in mind when he said the
empirical scientist is permitted to ignore the implications of the
transcendental perspective by treating objects as things in themselves [see
P3:136].
Kant's first Critique is often regarded as a book on the philosophy of
science--in particular, a book devoted to working out the epistemological
underpinnings of Euclidean geometry and Newtonian physics, in order to
demonstrate their absolute validity once and for all. I have argued in
this paper, however, that this is a gross oversimplification of his actual
intentions. Kant develops a "natural philosophy", a philosophy
devoted to working out the epistemological underpinnings of the ordinary
man's view of the natural world [K2:858-859]. To the extent that
science abandons this ordinary standpoint it passes beyond the bounds of
empirical knowledge and into the realm of hypothesis. Kant
consistently views his task as that of constructing a "propaedeutic
to the system of pure reason" [K2:xliii,25,869]. A "propaedeutic", he explains, is often "obtained
last of all, when the particular science under question has been already
brought to such completion that it requires only a few finishing touches to
correct and perfect it" [K2:76]. In this sense, his system is a kind
of philosophy of science on behalf of Euclid and Newton. (Indeed, it is
no accident that the title given to the main part of the Critique,
"Transcendental Doctrine of Elements", mimics the title of
Euclid's classic book.16) However, a
"propaedeutic" is also
forward-looking: in Kant's case it points to the "metaphysics"
of the future [K2:878]. (Thus, the title of the companion volume to K2 is
Prolegomena to Any Future Metaphysics...). Moreover, as we
have seen, by demonstrating that the classical scientific systems depend on and
arise out of the employment of clearly defined transcendental limits,
Kant's system presents science with a series of open questions as to how far
other, hypothetical models can be developed and applied to the world we
experience.
Since 1781 there have been four major revolutions which are of particular
interest to philosophers of science: the development of non-Euclidean
geometries by Lobachevsky in 1840 and Riemann in
1854, together with their subsequent application as explanations of the
structure of real space; the foundations set by Frege
in 1879 for the development of sophisticated, extra-Aristotelian logical
systems; the innovation of relativity physics by Einstein in 1905; and the
elaboration of quantum mechanics by Bohr and Heisenberg in 1927. For too
long these surprising revolutions have been regarded as evidence against
a Kantian interpretation of the world. Yet if the foregoing
interpretation of Kant is correct, then it is high time we come to
realize--especially in light of the current trends in the philosophy of
science, the trends away from any attempt to give definitive statements as to
what science is, and towards a more skeptical, historical/methodological
approach--that the world of modern science is a thoroughly Kantian
world, and that this is no coincidence! The time has come, that is, to
stop asking whether Kant's philosophy can still be valid in light of modern
developments, and to begin asking instead, whether these developments would
have arisen had Kant not explicated the world view on which they have their
philosophical foundation. My intention in this article has been to
demonstrate that, in fact, the first of these revolutions is consistent with,
if not based upon, the very Copernican turn which Kant's System
established. Further investigation would reveal that in each case these
new developments are based on the same principle, the principle which Kant
used to structure the fabric of his entire System, the principle of perspective.
Stephen R. Palmquist
Hong Kong Baptist College
BIBLIOGRAPHY
A1: Henry Allison,
The Kant-Eberhard Controversy (London:
The Johns Hopkins University Press, 1973).
B1: Gerd Buchdahl, "Kant's
'Special Metaphysics' and the Metaphysical Foundations of Natural Science"
in R.E. Butts (ed.), Kant's Philosophy of Physical Science (D. Reidel Publishing Company, 1986), pp.127-161.
C1: Paul M. Churchland, Matter and Consciousness2
(Cambridge, Mass.: The MIT Press, 1988 [1984]).
C2: Morris R.
Cohen, A Preface to Logic (London: George Routledge
& Sons Ltd, 1946).
E1: James
Ellington, "Translator's Introduction" to Immanuel Kant's Metaphysical
Foundations of Natural Science (New York: The Bobbs-Merrill
Company, Inc., 1970), pp.ix-xxxi.
E2:
A.C. Ewing, A Short Commentary on Kant's Critique of Pure Reason
(London: Methuen & Co. Ltd, 1950 [1938]).
F1: J. Fang, Kant-Interpretationen I (MÅnster:
Verlag Regensberg, 1967).
F2: Michael
Friedman, "Kant's Theory of Geometry", The Philosophical Review
94.4 (October 1985), pp. 455-506.
H1: Stephen
Hawking, A Brief History of Time (London: Bantam Books, 1988).
H2: Martin
Heidegger, Kant and the Problem of Metaphysics, tr. J.S. Churchill
(London: Indiana University Press, 1962).
H3: J. Hopkins, "Visual Geometry", The
Philosophical Review 82.1
(January 1973), pp.3-34.
K1: Immanuel Kant,
Thoughts on the True Estimation of Living Forces, excerpts translated by
N. Kemp Smith in Kant's Inaugural Dissertation and Early Writings on Space
(London: The Open Court Publishing Company, 1929).
K2: -----, Critique of Pure Reason,
tr. Norman Kemp Smith
(London: The
Macmillan Press Ltd., 1933 [1929]).
K3: -----, Critique
of Practical Reason, tr. Lewis White Beck (Indianapolis: The Bobbs-Merrill Company, Inc., 1956).
K4: -----, Critique
of Judgment, tr. James Creed Meredith (Oxford: The Clarendon Press,
1952).
M1: Gottfried
Martin, Kant's Metaphysics and Theory of Science, tr. P.G. Lucas
(Manchester: Manchester University Press, 1955).
M2: Arthur Melnick, "Kant's Theory of Space as a Form of
Intuition", in Richard Kennington (ed.), The
Philosophy of Immanuel Kant (Washington D.C.: The Catholic University
of America Press, 1985), pp.39-56.
P1: Stephen Palmquist, "Faith as Kant's Key to the Justification
of Transcendental Reflection", The Heythrop
Journal 25.4 (October 1984), pp.442-455.
P2: -----,
"The Radical Unknowability of Kant's "Thing
in Itself"', Cogito 3.2 (March 1985), pp.101-115.
P3: -----, "Six Perspectives on the
Object in Kant's Theory of
Knowledge", Dialectica 40.2 (1986), pp.121-151.
P4: -----, "The Architectonic Form of
Kant's Copernican Logic",
Metaphilosophy
17.4 (October 1986), pp.266-288.
P5: -----, "Is Duty Kant's 'Motive' for Moral
Action?", Ratio 28.2
(December 1986),
pp.168-174.
P6: -----, "A Kantian Critique of Polanyi's 'Post-Critical
Philosophy', Convivium 26 (March 1987), pp.1-11.
P7: -----,
"Knowledge and Experience -- An Examination of the Four Reflective 'Perspectives'
in Kant's Critical Philosophy", Kant-Studien
78.2 (1987), pp.170-200.
P8: -----, "Kant's Cosmogony
Re-Evaluated", Studies in History and
Philosophy of Science 18.3 (September 1987), pp.255-269.
P9: -----, "A
Priori Knowledge in Perspective: Mathematics, Method, and Pure
Intuition", The Review of Metaphysics 41.1 (September 1987),
3-22.
P10: -----, "A
Priori Knowledge in Perspective: Naming, Necessity and the Analytic A
Posteriori", The Review of Metaphysics 41.2 (December 1987),
pp.255-282.
P11: H.J. Paton, Kant's
Metaphysic of Experience, vol.1 (London:
George Allen & Unwin Ltd, 1936).
P12: Robert B. Pippin, Kant's Theory of Form
(New Haven: Yale
University Press, 1982).
R1: Hans Reichenbach, The Philosophy of Space and Time, tr.
M. Reichenback and J. Freund (New York: Dover
Publications, Inc., 1958).
R2: Bertrand
Russell, History of Western Philosophy (London: George Allen &
Unwin Ltd, 1946).
R3: -----, Introduction
to Mathematical Philosophy (New York: Simon & Schuster, n.d.).
S1: P.F. Strawson, The Bounds of Sense (London: Methuen
& Co. Ltd, 1966).
S2: Richard Swinburne,
Faith and Reason (London: The Clarendon
Press, 1981).
W1: Robert Paul
Wolff, Kant's Theory of Mental Activity (Cambridge, Mass.: Harvard
University Press, 1963).
FOOTNOTES
1. All references will be
included in the text using the abbreviations given in the Bibliography.
References to K2 (Kant's first Critique) will follow the second edition,
except where the material is unique to the first edition, in which case the
page number will be prefixed with "A". References to Kant's
other works cite the pagination of the Berlin Academy edition of Kant's works.
2. S2:83. Churchland makes a similar claim in C1:84: "Both
Euclidean Geometry and Newtonian physics have since [Kant's day] turned out to
be empirically false, which certainly undermines the specifics of Kant's
story." According to Paton, "The
possibility of new mathematical concepts is certainly not excluded by Kant's
theory" [P11:161]; however, in his opinion, "Kant's doctrine [of
space and time] is altogether too simple in the light of modern
discoveries" [163]. Nevertheless, he thinks we can at least admit
that Kant's theory of the synthetic a priori nature of geometrical propositions
is consistent with Euclid's methods [159]. Ewing is more
optimistic. He says in E2:47 that Kant's view of Euclidean geometry
"must not be dismissed at once as necessarily undeserving of any further
consideration", for even "Einstein himself admits that space may be
Euclidean" [46]!
Fortunately, some Kant-scholars do recognize that the overthrow of classical
scientific paradigms does not require the rejection of Kant's Critical
System. This "old-fashioned charge", as Buchdahl
puts it, "is wide of the mark" [B1:146]. At worst, he adds, the
modern revolutions indicate only that the Critical System's "'application'
can no longer take place in the way envisaged by Kant, and that the material to
which it was 'applied' stands in need of conceptual revision" [156].
He maintains that Kant intended his System to "function as a series of
constraints on the choice of possible hypotheses and of possible theoretical
formulations" [146], of which Newtonian physics is but one type.
This view of Kant as proposing a highly "progressive" philosophy of
science is supported by Fang [F1:106], who explicitly states (but unfortunately,
leaves undeveloped) that K2 is "perfectly compatible with the philosophy
of Relativity Theory and Quantum Mechanics, to say nothing of modern
mathematics" [123]. Ellington defends the same claim in a bit more
detail in E1:xxx.
3. The foregoing perspectival interpretation is used in P4 to explicate the
formal structure of Kant's System. The notion of a perspective is
defined, and its application to Kant's four main perspectives is demonstrated,
in P7. The perspectival way of interpreting
Kant is applied to his theory of the object of knowledge (e.g. terms such as
"thing in itself" and "appearance") in P3. And it is
applied to his moral philosophy in P5 (from which is taken much of the material
in the current paragraph in the main text). In several other articles I
have used this way of interpreting Kant to answer criticisms of Kant made by
Michael Polanyi [in P6], Philip Kitcher
[in P9] and Saul Kripke [in P10]. The justification
of the entire Critical enterprise, with its questionable assumption of the
"thing in itself", is defended on perspectival
grounds in P1 and P2. Finally, I have discussed Kant's general attitude
towards science (or "natural philosophy") in P8.
The summary of my interpretive framework given here differs in two respects
from my previous publications. First, I now refer to the third standpoint
(previously the "empirical") as "judicial" in order to
avoid confusing it with the empirical perspective within each standpoint
(a potential confusion Kant himself recognizes in K4:178-179). Although
this standpoint is "empirical" in the sense that it does deal with
particular aspects of man's experience (e.g. aesthetic or religious
aspects) much more fully than in reasoning based on the theoretical or
practical standpoints, this use of "empirical" must be carefully
distinguished from the use in the important transcendental-empirical
distinction. Moreover, in light of K2:739 ("There is no need of a
critique of reason in its empirical employment") the label
"judicial" (i.e., relative to judgment) more adequately reflects the
transcendental status of this standpoint, as well as the fact that its scope is
broader than the empirical perspective within each system.
The second change is that I now refer to the fourth (previously
"practical") perspective in each of Kant's three systems as the
"hypothetical" perspective. This avoids the potential confusion
between it and the practical standpoint, as adopted in K2. My use
of "practical" was potentially misleading because Kant normally
equates "practical" with "moral", whereas the fourth
perspective of the theoretical and judicial standpoints is not limited to
morality. The word "hypothetical" is an appropriate replacement
because it suggests the "as if" character of all conclusions
established from this perspective. (This is made especially obvious in
the Dialectic of K2.)
4. Thus Gottfried Martin
suggests in M1:96 that a helpful way of describing Kant's "a priori"
approach to science is to see it as a way of "considering Aristotelian
logic as the foundation of all logics, Euclidean space as the foundation of all
other spaces, and classical physics as the foundation of all other kinds of
physics." Unfortunately, Martin does not fully develop the
implications of this suggestion. It describes quite succinctly, however,
the interpretation of Kant's intentions which I am defending in this paper,
according to which a priori science differs in kind from a posteriori
sciences inasmuch as the latter depends on adopting a transcendental rather
than an empirical perspective.
5. M1:24. Martin
explains that the search for non-Euclidean geometrical systems was well under
way by Kant's day [M1:17-18]. In fact, Kant's friend Lambert practised an early form of non-Euclidean geometry, so it
is likely that Kant was well aware of these developments.
Although Friedman (rightly) thinks Kant believes "there is no way to draw,
and thus no way to represent, a non-Euclidean straight line" [see note 9
below], he (wrongly) concludes "the very idea of a non-Euclidean geometry
is quite impossible" for Kant [F2:488]. He never says what he means
by "idea", yet he uses the word in similar ways several times:
"pure intuition [for Kant] cannot be said to provide a model for Euclidean
geometry at all; rather it provides the one and only possibility for a rigorous
and rational idea of space" [505; see also 504]. Unless
Friedman has a very unusual (and unKantian) meaning
for "idea", such claims are quite obviously unfounded. Indeed,
I will argue in section IV that Kant's theory of "ideas" in the
Dialectic is precisely the proper place to locate the possibility of
non-Euclidean geometries in a modern reconstruction of Kant's System.
In F2:502-503 Friedman claims Kant "has no notion of possibility on which
both Euclidean and non-Euclidean geometries are possible." To defend
his view he refers to various texts in the Postulates [K2:265-294], where Kant
is intentionally limiting his attention to real possibility.
Friedman describes Kant's position with admirable clarity when he mentions
"two notions of possibility: 'logical possibility,' given by the
conditions of thought alone; and 'real possibility,' given by the conditions of
thought plus intuition" [F2:503]. However, he then makes the
outrageous claim: "this line of thought employs a notion of logical
possibility that is completely foreign to Kant." In the Dialectic,
and throughout Kant's Critical System, Kant's distinction between logical and
real possibility plays an essential role in his arguments [see e.g.
K2:xxviin]. (This point is forcefully argued by Allison in A1.)
Friedman's mistake is to limit his understanding of Kant's view of possibility
to the strict comments regarding the limitations of real possibility
which he makes in the Postulates. (In any case, such "real"
possibility refers not to what is physically possible, but to what can
be constructed or represented by the human perceiver.) That
Kant believes non-Euclidean geometries are logically possible is made
unambiguously clear in K2:268: "there is no contradiction in the
concept of a figure which is enclosed within two straight lines". A
better way of expressing the point Friedman is trying to make, therefore, would
be to say Kant downplays the importance of merely logical possibility.
6. "The apodeictic certainty of all geometrical propositions"
is also exemplified in K2:A24 by referring to the Euclidean doctrine "that
there should be only one straight line between two points".
7. S1:277. Martin
demonstrates, by contrast, that in K2:38 alone there are two carefully structured
arguments for the transcendental ideality of space which do not refer at all to
geometry: indeed, he traces the significance of these arguments back to
Plato and Aristotle [M1:29f,32f]. Moreover, Pippin not only argues
explicitly against Strawson's claim, but provides a
list of other recent works which do the same [P12:55n; see also 73-74n]. Strawson's interpretation is criticized most thoroughly in
H3.
8. I discuss the proper interpretation
of Kant's doctrine of "pure intuition" in P9 [see also note 11
below]. The typical misinterpretation is made by Melnick
in M2:51-54 when he refers to pure intuition as "an imaginative act
or sequence" by which something is constructed. As long as
"imagination" is understood in its ordinary sense (as a kind of
internal perception) rather than in Kant's special, transcendental sense (as a
kind of pre-conscious requirement for perception), such a description of
Kant's theory of pure intuitive construction is sure to be misleading. In
K2:299 Kant clearly explains that, because the basic principles of Euclidean
geometry are pure intuitions, they are essentially non-sensible
("generated in the mind completely a priori"); only if the
mathematician can construct them in empirical intuition, as
"an appearance present to the senses", can we actually experience
them, even though they are originally "produced a priori"
in the transcendental recesses of our mind.
9. There is some
potential for misunderstanding on this matter of the picturability
of non-Euclidean geometrical figures. Hawking defends the established
position in H1:24: "It is impossible to imagine a
four-dimensional space." Nevertheless, Martin argues "that
there are ways in which non-Euclidean geometries can also be constructed, by
purely analytical means or by constructing Euclidean models of non-Euclidean
geometries" [M1:25]. Such "construction", however, is logical
in the former case, and Euclidean in the latter, so neither type
provides actual non-Euclidean pictures. Indeed, it seems clear enough
that, whenever we make pictures of higher-dimensional spaces, we always do so
by adding specially defined lines onto Euclidean pictures. Even simple non-Euclidean
principles, such as the assertion that two different straight lines can pass
through two points, can be pictured only in a Euclidean way, such as by drawing
at least one of the lines as curved, and pretending it is
straight. Thus, Reichenbach, despite his thorough
discussion of the picturability of non-Euclidean
geometries in R1:37-92, agrees that a non-Euclidean geometrical proposition
"contradicts the human power of visualization" [3].
10. See my defence of this claim in P3:136.
11. See P3:130-134.
In opposition to Kant, Russell argues that there are only two kinds of geometry
now thought to be valid: "pure geometry" and "geometry as
a branch of physics... Thus of the two kinds of geometry one is a
priori but not synthetic, while the other is synthetic but not a priori"
[R2:743]. This view, which entails that the Euclidean use of pictures is
obsolete and can be--indeed, ought to be--replaced by purely analytic, logical
representations, is widely accepted today [see e.g. C2:9-11 and
F2:455-457]. Thus in R3:145 Russell chides Kant for inventing "a
theory of mathematical reasoning according to which the inference is never
strictly logical, but always requires...'intuition'. The whole trend of
modern mathematics...has been against this Kantian theory."
Strawson calls Russell's twofold distinction
"the positivist view" [S1:278f] and argues that "a third
way" is to restrict geometry to an analysis of the way things appear
[286]. This way of viewing Euclidean geometry is the only respect in
which its propositions can be called self-evident, and is important especially
"in the initial stages of learning geometry" [286].
Moreover, Strawson rightly sees that, as we shall see
in section III, this "phenomenal geometry" plays an important role in
the empirical application of Kant's theory [287]. The main danger in Strawson's view is a tendency to ignore Kant's distinction
between a picturable "image" and a rational
"schema": the latter "can exist nowhere but in
thought. It is a rule of synthesis of the imagination, in respect to pure
figures in space" [K2:180]. So the picturable
(Euclidean) image becomes applicable only through such a schema, which is in
itself not a visual representation. Thus, what Strawson
calls "phenomenal geometry" is the empirical effect of the
transcendental operations of pure intuition. As such, the visual images
themselves should be regarded as epiphenomena of Kant's theory.
Friedman contrasts Russell's view with what he calls the "anti-Russellian interpretation", which views Kant as
agreeing with Russell that mathematical reasoning is analytic, but as viewing
the axioms on which such reasoning is based as synthetic [F2:486-489;
see also 502]. After examining Kant's views on the empirical procedure
of mathematical proof, which he mistakes for the entirety of Kant's view,
Friedman confidently concludes that this alternative is "rather obviously
untenable and definitely unKantian" [498].
The reason, he explains, is that arithmetic, for Kant, "differs from
geometry precisely in having no axioms" [490]. "For
Kant...arithmetical propositions are established by calculation, a
procedure that is sharply distinguished from logical argument in being
essentially temporal" [491]. This "distinction between calculation
and logical argument", he suggests, "is perhaps most basic to Kant's
conception of the role of intuition in mathematics" [492].
Friedman's position is seriously inadequate, however, in several
respects. First, aside from a brief comment in a footnote [F2:49n], he
never defends his position against the opposite view, defended at length by
Martin, "that Kant discovered the axioms of arithmetic...and that the
axiomatic theory of arithmetic starts from these Kantian discoveries"
[M1:18]. Secondly, his belief that the synthetic character of mathematics
is related for Kant to the temporal character of calculation reveals quite a
naive interpretation of "pure intuition" (much like that of Kitcher, which I thoroughly criticize in P9). Thus he
opines, "the whole point of pure intuition [for Kant] is to enable us to avoid
rules...by actually constructing the desired instances" [469]. Such
geometrical construction (like all arithmetical calculation, or for that
matter, all logical argumentation), however, is thoroughly an a posteriori
activity, and misses the whole point of Kant's emphasis on the a priori [see
P10, esp. section IV]. Friedman quotes K2:xi-xii in defence
of his conclusion about the reasoning process [F2:501]. Yet he
reads quite an unKantian meaning into the word
"method". "The true method" for demonstrating
"the properties of the isosceles triangle", Kant explains, is
"not to inspect what he discerned either in the figure [empirically, i.e.
synthetic a posteriori], or in the bare concept of it [logically, i.e. analytic
a priori]...; but to bring out what was necessarily implied in the concepts
that he had himself formed a priori..." [K2:xi-xii]. Friedman
thinks "method" here means "step-by-step procedure".
Yet Kant explains this "true method" by referring directly to a
"true perspective", or way of viewing the (empirical)
procedures involved in constructing such proofs. Thus there are grounds
for serious doubt concerning Friedman's confident conclusion: "I do
not see how there can be any doubt...that Kant's 'true method' of geometry is
precisely Euclid's procedure of construction with straight-edge and compass"
[F2:501]. On the contrary, nothing could be further from Kant's
mind! Kant is not suggesting that the philosopher should measure
metaphysics with a straight edge and compass (i.e. a posteriori)! Rather,
he is arguing that the philosopher, like the pure geometer, should adopt a
transcendental (synthetic a priori) perspective. Unfortunately, Friedman
(like Russell) seems to know alot more about geometry
than he does about the purposes and methods of Kant's Critical System.
12. Friedman admits that
K2:155n implies a distinction between pure and applied mathematics [F2:482n],
yet he asserts with little actual explanation or argument [see note 15 below]
that Kant had nothing corresponding to "our modern distinction
between pure and applied mathematics" [F2:504n]. My interpretation
of K2:155n implies that Friedman's view is quite mistaken. Kant's most
important perspectival distinction--between the
transcendental and empirical perspectives--can be regarded as a forerunner of
this modern distinction. The differences which disturb Friedman should
not be described in terms of "modern" versus "out-of-date",
but rather in terms of "mathematical" versus
"philosophical". Moreover, Martin explains, in defending the
traditional interpretation, that Kant's goal was to solve the problem of
"why Euclidean geometry, in spite of being a pure construction in thought,
is valid for Newtonian physics" [M1:36]. "By discovering the
connection between mathematics and natural science Kant fundamentally went far
beyond the [classical] concept of applied mathematics. What we have is
not a ready-made mathematics which is applied in physics, but one fundamental
human faculty which is active in mathematics and in physics" [36]. And
this is much closer to the modern ways of viewing this problem than Friedman
acknowledges.
13. This standpoint is
reflected in an interesting way in Heidegger's interpretation, when he says
"the ego cannot be conceived [by Kant] as temporal, i.e. as
intra-temporal, precisely because the self originally and in its innermost
essence is time itself" [H2:200-201].
The close correlation between space and time as forms of human intuition
suggests that there ought to be a synthetic a priori science of time,
corresponding to transcendental (Euclidean) geometry. Paton
cites the lack any clear reference to such a corresponding science in the
Aesthetic of K2 as a weakness of Kant's exposition [P11:128]. Wolff
argues in W1:230 that mechanics is to time as geometry is to
space. Most interpreters, however, assume arithmetic to be the
science in question, since Kant uses an example from arithmetic in K2:15-16
just before introducing his example from geometry, both of which are meant to
clarify the nature of synthetic a priori judgments. (He also pairs
together the "propositions of arithmetic or geometry" in
K2:764.) Friedman even claims that arithmetic is the primary
science for Kant, since inner sense is prior to outer sense [F2:495-496].
But this question of priority is debatable, since Kant does discuss space and
outer sense before time and inner sense in the Aesthetic, and because
space is given a measure of priority over time in the Refutation of Idealism
[K2:274-279], where Kant argues that our consciousness in time presupposes
something permanent in space. My view is that from the transcendental
perspective, time has priority, whereas from the empirical perspective, space
has priority.
14. This interpretation of
the correlation between the sections of K2 and the perspectives Kant adopts is
defended in P4 and P7, and in more detail in my forthcoming book, Kant's
System of Perspectives, chs 2-4,7.
15. The only redeeming
aspect of Friedman's otherwise abstruse treatment of Kant's theory of geometry
[see notes 5, 11, 12] is that he correctly stresses the close relationship
between Kant's attitudes towards Euclidean geometry and Aristotelian logic
[F2:456]. Unfortunately, after pointing out that "our logic, unlike
Kant's is polyadic [quantifier-dependent]
rather than monadic (syllogistic)" [460], he infers that Kant was
simply wrong, though he adds that we should not blame him too harshly. His
argument runs something like this: (1) our modern view of logic and
mathematics began with Frege in 1879; (2) Kant
developed his Critical philosophy a century before Frege's
revolution; therefore (3) Kant cannot have suggested or had in mind anything
like our modern distinctions [see e.g. F2:481-483,488-489,500n,503n]. In
a sequel to the present article I plan to argue, by contrast, that Kant laid
the groundwork for, and in some respects anticipated, the modern innovations to
which Friedman refers [see note 16 below]. The reason this is not often
recognized is that the foundations he laid were thoroughly philosophical,
not mathematical.
16. Moreover, just as
Kant's "Elements" begins with an exposition on how to view Euclid's Elements
in its proper perspective, so also it continues by doing the same with three
other classical traditions. The "Analytic of Concepts" develops
a doctrine of categories which enables us to view Aristotle's classic
book, Categories, in its proper perspective. The "Analytic of
Principles" develops a doctrine of principles which enables us to
view Newton's classic book, Principia, in its proper perspective.
And the "Dialectic" develops a doctrine of ideas which enables
us to view the corresponding classical theory of Plato (as expressed in his Dialogues)
in its proper perspective. The first Critique is an attempt to
show that each of these classical theories has a proper role to play in a
complete system of theoretical philosophy, but that the validity of each can be
established only by limiting each to a specific perspective. Without a
clear recognition of Kant's Critical attitude towards his tradition (an
attitude which always includes both negative and positive aspects), the significance
of this modern classic cannot be fully understood.
This etext is based on a prepublication draft of the published
version of this essay.
Send comments to: StevePq@hkbu.edu.hk
My Web Counter identifies you as visitor number
to this page, last
updated on 29 March 2011. Please come again!
Back to the listing
of Steve Palmquist's published articles.
Back to the main map of Steve Palmquist's web site.