Anda di halaman 1dari 19


Michael Friedman and Graham Bird
IMichael Friedman
This paper considers the extent to which Kants vision of a
distinctively transcendental task for philosophy is essentially tied to his views
on the foundations of the mathematical and physical sciences. Contemporary
philosophers with broadly Kantian sympathies have attempted to reinterpret his
project so as to isolate a more general philosophical core not so closely tied to the
details of now outmoded mathematical-physical theories (Euclidean geometry and
Newtonian physics). I consider two such attempts, those of Strawson and
McDowell, and argue that they fundamentally distort the original Kantian impulse.
I then consider Buchdahls attempt to preserve the link between Kantian
philosophy and the sciences while simultaneously generalizing Kants doctrines
in light of later scientific developments. I argue that Buchdahls view, while not
adequate as in interpretation of Kant in his own eighteenth century context, is
nonetheless suggestive of an historicized and relativized revision of Kantianism
that can do justice to both Kants original philosophical impulse and the radical
changes in the sciences that have occurred since Kants day.


ant begins the Critique of Pure Reason with the distinction

between empirical knowledge and pure or a priori
knowledge. He points out that metaphysics has always aimed at
the latter type of knowledge and, indeed, at knowledge that
extends the sphere of our judgements beyond all limits of
experience (A3/B6).1 Metaphysics has here taken inspiration
from the example of mathematics, but in doing so, Kant suggests,
metaphysics has also been fundamentally misled. Whereas the
latter science has been so far occupied chiefly with analysis, with
what has already been contained in our concepts (although
confusedly), the former science essentially goes beyond mere
concepts, in that it occupies itself with objects and cognitions
solely in so far as they can be presented in intuition (A46/B8
9). And since only the latter type of a priori knowledge, synthetic
as opposed to analytic a priori knowledge, genuinely extends the
range of our cognition, metaphysics, if it, too, hopes to extend the
1. All references to the Critique of Pure Reason are given parenthetically in the text by the
standard pagination of the first (A) and second (B) editions respectively.



sphere of our judgements, requires a preliminary inquiry into the

nature, scope, and limits of our synthetic a priori knowledge. The
preliminary inquiry in question is then precisely the task of the
Critique, and only in this way, Kant holds, can philosophy or
metaphysics finally enter into the secure path of a science.
With this articulation of the peculiar task of what Kant now calls
transcendental philosophy, the subject matter of philosophy has,
for the first time, been clearly delimited from that of the natural
and mathematical sciences, and, in particular, from those elements
of a priori knowledge present in the latter sciences themselves:
[N]ot every a priori cognition should be called transcendental, but
only that through which we know that and how certain representations (intuitions or concepts) are applied wholly a priori, or are
possible (that is, [through which we know] the possibility or the a
priori employment of the cognition). Therefore, neither space nor
any a priori geometrical determination thereof is a transcendental
representation, but what can alone be called transcendental is the
knowledge that these representations are not at all of empirical
origin, and the possibility that they can nevertheless relate a priori
to objects of experience. (A56/B80)
I term all cognition transcendental which occupies itself in general,
not so much with objects, but rather with our mode of cognition of
objects, in so far as this is supposed to be possible a priori. (B25)

Transcendental philosophy is thus a meta-discipline, as it were,

whose distinctive task is to investigate the nature and conditions
of possibility of first-level scientific knowledge.
Contemporary philosophers dissatisfied with the prevailing
tendency to naturalize their discipline by incorporating philosophy, too, among the natural sciences are therefore understandably drawn back to the philosophy of Kant, who thus first
staked out the claim to a distinctive, transcendental task lying
outside the domain of the first-level sciences themselves. When it
comes to the details of Kants own transcendental program,
however, such contemporary philosophers immediately run into
apparently insuperable difficulties. For Kant makes it very clear
that his inquiry breaks down, more specifically, into the two
questions How is pure mathematics possible? and How is pure
natural science possible?where the first concerns, above all, the
possibility of Euclidean geometry, and the second concerns the
possibility of fundamental laws of Newtonian physics such as



conservation of mass, inertia, and equality of action and reaction

(B1921). Now, however, we no longer believe that these specific
Kantian examples of synthetic a priori knowledge are even true,
much less that they are necessarily and a priori true. Indeed, for
precisely this reason, we are no longer convinced that there are any
real examples of synthetic a priori knowledge at all. So it seems
that we cannot now follow Kant in even his very first step in further
articulating what the distinctively transcendental task of philosophy is supposed to be.
It is also understandable, then, if contemporary philosophers
with broadly Kantian sympathies attempt to reinterpret his project
so as to isolate a more general philosophical core not so closely
tied to the details of now outmoded mathematical-physical
theories. And it is similarly understandable, further, if such
philosophers attempt to break the intimate link between Kantian
transcendental philosophy and problems in the foundations of
mathematics and physics by reorienting Kants transcendental
argumentation towards much more general philosophical
problems, which, it is hoped, can be pursued largely independently.
In particular, by thus separating a more general core from its more
vulnerable accompanying parts, now seen as inessential, we might
hope to rehabilitate Kants claim to a distinctive task for
philosophy in the face of the contrary ambitions of contemporary
philosophical naturalism.

No-one within the analytic tradition has done more to reawaken
interest in Kantian transcendental philosophy than P. F. Strawson.
In his classic essay, The Bounds of Sense, Strawson sets out to
effect a separation or division between what remains fruitful and
interesting [in Kants Critique] and what no longer appears
acceptable, or even promising, in its doctrines.2 Among the
obstacles to sympathetic understanding, of course, is the state
of scientific knowledge at the time at which Kant wrote, which
inclined him, in particular, to a belief in the finality of Euclidean
geometry, Newtonian physics, and Aristotelian logic.3 Strawsons
2. P. F. Strawson, The Bounds of Sense (London: Methuen, 1966), p. 16.
3. Op. cit., p. 23.



aim, accordingly, is to abstract out a more general core of Kantian

doctrine, which, in his own words, attempts to articulate the
necessary conditions of the possibility of experience in general
rather than simply the necessary presuppositions of physical
science.4 These necessary conditions of the possibility of
experience in general, freed of any dependence on the specifics
of scientific theorizing, will then be independent of the changes
and vicissitudes that such theories have actually undergone in the
subsequent history of the sciences.
The defensible core of Kantian doctrine is thus devoted to what
Strawson calls the metaphysics of experience, where Kant
attempts to show what the limiting features must be of any notion
of experience which we can make intelligible to ourselves.5
Prominent among these limiting features are that, in virtue of what
is required for the possibility of self-consciousness, experience
must include awareness of objects which are distinguishable from
experiences of them, that, as a further consequence, there must
be one unified (spatio-temporal) framework of empirical reality
embracing all experience and its objects, and, finally, that certain
principles of permanence and causality must be satisfied in the
physical or objective world of things in space.6 The spatiotemporal framework in question, however, is not further limited by
the constraints of any particular mathematical-physical theory: it
is not limited to specifically Euclidean space, for example, or to
the structure of Newtonian, as opposed to relativistic, physics. Nor
are the principles of permanence and causality to which Strawson
refers identical to the strong Newtonian principles of conservation
and thoroughgoing causal determination that Kant actually
defends; on the contrary, we require only relatively, as opposed to
absolutely permanent re-identifiable objective items subject to
more or less regular law-like expectations.7
In this way, far from being limited to describing the foundations
or necessary presuppositions of the mathematical and physical
sciences of Kants day, the defensible core of Kants transcendental argumentation has a much more general philosophical

Op. cit., p. 121.

Op. cit., p. 24.
Op. cit., p. 146.



aim: namely, to show that the impoverished conception of

experience of the classical empiricist sense-datum theorist,
according to which our experience is confined, at least initially, to
the veil of perception of our own subjective sense impressions, is
not in fact intelligible. And it is precisely this conception of
experience, of course, which sets the stage for the classical
sceptical arguments of the modern period. Such arguments, if the
defensible core of Kants transcendental argumentation is correct,
operate with an entirely incoherent conception of an initial, purely
subjective experience, which, in particular, fails to make room
even for the self-conscious subject that is supposed to be the bearer
of this experience. Therefore, if the argumentation in question is
correct, we will have made decisive progress in the defense of our
commonsense conception of experience and the world from the
strictures of philosophical scepticism.
Now this Strawsonian reconstruction of the defensible core of
the Critique has spawned an extensive literature on the precise
anti-sceptical force of what are now known as transcendental
arguments. Perhaps the most important point in this context has
been made in a well-known paper by Barry Stroud,8 according to
which the most such arguments can prove is that our overall
conceptual scheme must possess certain features (here, the
conception of an external world subject to constraints of
permanence and regularity) on pain of incoherence. They do not
and cannot show, however, that such features and conceptions must
actually be correct (that there really is in external world
conforming to these constraints, for example), and, in this sense,
their anti-sceptical force is severely limited. Strawson, in
responding to Strouds challenge, has fully acknowledged this
point: the role of such transcendental arguments is solely to show
that one type of exercise of conceptual capacity is a necessary
condition of another, so that their precise philosophical force is
simply to articulate necessary conceptual connections within our
conceptual scheme. The peculiarly philosophical task of
Strawsonian descriptive metaphysics, then, is to exhibit the
necessary structure of our conceptual scheme, which is thereby
8. B. Stroud, Transcendental Arguments, The Journal of Philosophy, Vol. LXV, No. 9,
May 1968.



depicted as a coherent whole whose parts are mutually supportive

and mutually dependent, interlocking in an intelligible way.9
What, more precisely, is the nature of the necessary conceptual
connections we are thus attempting to delineate, and what, more
specifically, is the character of the peculiarly philosophical
enterprise wherein such delineation takes place? In The Bounds of
Sense Strawson firmly rejects both Kants conception of synthetic
a priori truth and his conception of the transcendental philosophical
task of explaining the possibility of such truth. But to the question
of what alternative conception might be offered, Strawson is
remarkably noncommittal:
To this I can only reply that I see no reason why any high doctrine
at all should be necessary here. The set of ideas, or schemes of
thought, employed by human beings reflect, of course, their nature,
their needs and their situation. They are not static schemes, but
allow of that indefinite refinement, correction, and extension which
accompany the advance of science and the development of social
forms.... But it is no matter for wonder if conceivable variations are
intelligible only within a certain fundamental general framework
of ideas, if further developments are conceivable only as
developments of, or from, a certain general basis.10

The question as to the precise nature of the peculiarly philosophical enterprise, wherein we attempt to articulate what the
limiting features must be of any notion of experience which we
can make intelligible to ourselves, is here rather abruptly
In Skepticism and Naturalism, however, Strawson returns to this
question, and in a way, I believe, that is particularly revealing.
Prompted largely by Quines attack on the very notion of a special
realm of necessary conceptual connections, Strawson
acknowledges that there is indeed a philosophical problem about
the character of abstract and general thinking which at least
appears to concern itself directly with concepts or universals, as
in philosophy itself, or with other abstract objects, as in
mathematics.11 The kind of hard naturalism represented by
Quine rejects the existence of a peculiar domain of objective
9. P. F. Strawson, Skepticism and Naturalism (New York: Columbia, 1985), pp. 2223.
10. The Bounds of Sense, p. 44.
11. Skepticism and Naturalism, p. 86.



mathematical or conceptual truth as falling entirely outside the

purview of modern empirical natural science. But it is also possible
to endorse a soft naturalism, Strawson urges, in which we simply
accept our apparent capacity for insight into such truths as sui
generis, thereby falling in with the classical philosophical tradition
of rational intuition. We thereby endorse as Descartes would
say, the power of clear and distinct perception of necessary
truths; or, as both he and Spinoza would say, the power of rational
intuition of such truths.12
It is not my purpose to argue against this last Strawsonian
suggestion, but I do want to emphasize how profoundly unKantian
it is. For Strawson here aligns himself with the classical rationalist
tradition which Kant himself explicitly rejects. Indeed, from a
Kantian point of view, Strawson is here fundamentally uncritical
in several interrelated respects. In the first place, conceptual truth
in philosophy is intimately associated with mathematical truth,
even though, as Kant himself emphasizes, mathematics has long
since attained the secure path of a science, whereas philosophy
is very far from doing so (Bxxv). In the second place, necessary
or a priori truth in general is conceived as the product of a faculty
of clear and distinct perception or rational intuition, and thus
as secured by a special kind of quasi-perceptual acquaintance with
a particular domain of objects. The whole point of Kants
Copernican revolution in thinking about the a priori, by contrast,
is to break decisively with this classical tradition. Necessary or a
priori truths, in mathematics, for example, have no special domain
of objects at all; their characteristic status derives rather from their
constitutive function with respect to ordinary empirical objects
(viz., appearances): their function, namely, of making empirical
knowledge of such empirical objects first possible.13 Finally, and
correlatively, the peculiar role of philosophy, from a Kantian point
of view, rests neither on a special faculty of rational intuition nor
on the construction of a priori truths constituting the necessary
preconditions of empirical knowledge that we find in mathematics.
Philosophy functions rather on a higher or meta-level, as it were,
12. Op. cit., p. 91.
13. See especially 22 of the second edition Transcendental Deduction, and compare



where our task is precisely to explain how such first-level a priori

knowledge is possible.

A more recent attempt at importing central Kantian themes into
contemporary philosophical discussion is John McDowells Mind
and World,14 which, as McDowell points out, is very strongly
influenced by Strawsons reading of Kant in The Bounds of Sense.
Instead of focusing on the problem of philosophical scepticism,
however, and the Strawsonian project of delineating, in response,
the minimal necessary conditions of any intelligible experience,
McDowell concentrates rather on the fundamental Kantian
distinction between concepts and intuitions, understanding and
sensibility, spontaneity and receptivity. For McDowell, Kants
most important insight is that understanding and sensibility,
spontaneity and receptivity, must always be integrated together.
There is no room, in particular, for either unconceptualized
sensory input standing in no rational relation to conceptual
thought, or purely intellectual thought operating entirely
independently of all rational constraint from sense experience:
Thoughts without content are empty, intuitions without concepts
are blind (A51/B75). And it is only by fully assimilating this
Kantian insight, according to McDowell, that we can escape the
otherwise interminable philosophical dialectic or oscillation
between Coherentism and the Myth of the Given: the temptation,
on the one hand, to picture the understanding as a self-contained
conceptual sphere with no rational relation to an independent
empirical world, or, on the other, to invoke bare unconceptualized
sensory presences acting on the understanding from outside the
conceptual sphere.
In response to this philosophical dialectic, McDowell
recommends an alternative picture of sensible experiences,
products of our receptivity, as nonetheless thoroughly infused with
conceptual content: In experience one takes in, for instance sees,
that things are thus and so. That is the sort of thing one can also,
for instance, judge.15 In this way, experiences themselves are
14. J. McDowell, Mind and World (Cambridge: Harvard, 1994).
15. Op. cit., p. 9.



states or occurrences that inextricably combine receptivity and

spontaneity.16 Sense experiences belong to our spontaneity, the
sphere of the understanding, in virtue of their conceptual content:
their expressibility in a that-clause. Unlike acts of judgement,
however, sense experiences also belong to our receptivity, in that
we are passively presented with the worlds appearing to be thus
and so rather than actively judging (perhaps after reflectively
deciding whether to accept this appearance or not) that the world
is in fact thus and so. Hence, although passively received sensible
experiences belong, in virtue of their content, to the conceptual
realm, there remains a fundamental distinction between such
experiences, on the one side, and acts of judgement or belief, on
the other. We thereby retain the crucial idea of independent
constraint from outside the realm of our judgements and beliefs,
without falling into the temptation to invoke bare unconceptualized
sensory presences acting on the understanding from outside the
conceptual sphere.
McDowells most basic Kantian legacy, therefore, is the idea of
a necessary and thoroughgoing interdependence between the
spontaneity of the understanding and the receptivity of sensibility.
And it is just such a necessary interdependence that Kant himself
attempts to establish in the Transcendental Deduction of the
Categories, whose aim is precisely to show that [sensible]
appearances have a necessary relation to the understanding
(A119), so that the two extremes, namely sensibility and
understanding, must stand in necessary interconnection (A124).
McDowells own way of defending the necessary interconnection
in question, however, does not appeal to a transcendental
deduction. Instead, in accordance with a broadly Wittgensteinian
philosophical quietism, McDowell rather attempts to remove the
obstacles standing in the way of fully assimilating the Kantian
insight by providing us with a diagnosis of the intellectual
difficulties that actually give rise to the opposing philosophical
The key obstacle standing in the way of a proper appreciation
of the Kantian insight, according to McDowell, is the
disenchantment of nature effected by the scientific revolution.
Before the scientific revolution nature was seen as itself filled
16. Op. cit., p. 24.



with meaning, and thus as perfectly hospitable to the similarly

meaningful structure of the normative space of reasons
governing our understanding. After the scientific revolution,
however, nature was seen as a mere realm of law, which, in
particular, is thereby radically external to the meaningful
normative structure governing our understanding.17 Sensibility, by
contrast, clearly and incontrovertibly belongs to the domain of
nature; it belongs to our nature as perceiving organisms receiving
sensory impacts from the external natural world. But it then
becomes entirely mysterious, on the modern conception of
disenchanted nature, how sensibility and understanding can, after
all, stand in necessary interconnection, for it can seem impossible
to reconcile the fact that sentience belongs to our nature with the
thought that spontaneity might permeate our perceptual
experience itself.18 Yet this dilemma only arises in the first place,
McDowell continues, if we commit ourselves in advance to a bald
naturalism that simply equates the domain of nature as such with
the realm of law. And so, to relieve us of our intellectual burden,
McDowell recommends that we instead adopt an Aristotelian
relaxed naturalism, in which initiation into the space of reasons
is seen as a normal part of the maturation of adult human beings,
wherein natural processes such as sensory perception become
infused with conceptual meaning.19
Again, it is not my purpose here to challenge McDowells
diagnosis.20 As in the case of Strawsons reconstruction, however,
I do want to point out how profoundly unKantian it is. Indeed, there
is a more than superficial affinity between the Strawsonian
conception of an autonomous realm of necessary conceptual
connections and McDowells picture of an autonomous space of
reasons constituting the normative structure of our understanding.
In particular, both see a hard naturalism or bald naturalism
committed to the hegemony of the modern mathematical-physical
sciences as the primary threat to a proper appreciation of the philosophical autonomy they wish to defend, and both recommend, in
17. Op. cit., pp. 7072.
18. Op. cit., p. 70, and compare p. 108.
19. See op. cit., pp. 7684.
20. See, however, my Exorcising the Philosophical Tradition, The Philosophical Review,
Vol. CV, No. 4, October 1996.



response to this threat, a soft naturalism or relaxed naturalism

allowing us simply to embrace the autonomy in question as sui
generis. For both Strawson and McDowell, then, there is a
fundamental tension between the world-view of modern
mathematical-physical science, on the one side, and the Kantian
themes they wish to reintroduce into contemporary philosophy, on
the other. And it is clear in both cases, I believe, that the problem
in question arises directly out of the recent naturalistic attacks on
the autonomy of philosophy due, above all, to the work of Quine.
Kants own philosophical situation is entirely different, however,
and there is no such tension, for Kant, between his characteristic
conception of the philosophical enterprise and the world-view of
modern mathematical-physical science. On the contrary, the
primary aim of Kants Transcendental Aesthetic and Analytic, as
we saw at the beginning, is precisely to explain how pure
mathematics and pure natural science are possible, so as thereby
precisely to underwrite, in particular, the new conception of nature
articulated in the modern mathematical-physical sciences.
The best way to see the importance of this point, in the present
context, is to glance briefly at the crucial step in Kants own
argument for establishing the necessary interdependence of
understanding and sensibility. In this argument, the Transcendental
Deduction of the Categories, the necessary mediating role is
carried out by an activity or faculty Kant calls the transcendental
synthesis of the imagination or the pure productive imagination:
We thus have a pure faculty of imagination, as a fundamental faculty
of the human soul, lying a priori at the basis of all our cognition. By
means of it we bring the manifold of intuition, on the one hand, in
connection with the condition of the necessary unity of pure
apperception, on the other. The two extremes, namely sensibility
and understanding, must stand in necessary interconnection by
means of this transcendental function of the imagination, because
otherwise the former would indeed supply appearances, but no
objects of an empirical cognition, and thus no experience. (A124)

In 24 of the second edition version, entitled On the application of

the categories to objects of the senses in general, Kant explains that
the transcendental synthesis of the imagination is an action of the
understanding on sensibility, and its first application (at the same
time the ground of all the rest) to objects of an intuition possible for



us (B152). This transcendental synthesis, as belonging to the

productive rather than the reproductive imagination, is exerted on
pure rather than empirical intuition (A118, B150152), and Kant
proceeds to illustrate its action precisely by examples of
geometrical construction (the drawing of a straight line, the
describing of a circle), and also by the motion involved in such
geometrical construction, regarded as the external figurative
representation of time... whereby we attend to the succession of this
determination in [inner sense] (B154).
The pure imagination thus provides for both the application of
geometry to objects of the senses and the application of what Kant
calls the general doctrine of motion (B49). Kant himself insures
that sensibility and understanding stand in necessary
interconnection, in the first instance, by insuring that all empirical
objects conform to the structure of Euclidean geometry, and also,
if I am not mistaken, to the structure of what we now call
Newtonian spacetime. And we thereby guarantee, in precisely
this way, that synthetic a priori knowledge in fact serves as the
condition of possibility of all empirical knowledge. Far from
seeing the world-view of modern mathematical-physical science
as any kind of threat to his philosophical transcendental
argumentation in the Aesthetic and Analytic of the Critique, Kant
undertakes the key step in that argumentation precisely to secure
the philosophical foundations of this world-view once and for all.

From these examples, it appears that we cannot so easily abstract
Kants characteristic conception of transcendental philosophy
from the scientific context of his time. It appears, in particular, that
we cannot so easily leave aside Kants preoccupation with
synthetic a priori mathematical-physical knowledge without
distorting his own philosophical impulse entirely beyond
recognition. So it is especially significant, in this context, that there
have also been attempts within the analytic tradition to reconstruct
and rehabilitate Kants philosophy of science as well, perhaps the
most interesting of which, from the present point of view, is that
undertaken by Gerd Buchdahl.21 For one of Buchdahls principal
21. G. Buchdahl, Metaphysics and the Philosophy of Science (Oxford: Blackwell, 1969).



arguments is that we can preserve the centrality of mathematical

and physical scientific knowledge for Kantian philosophy, while,
at the same time, loosening its apparent commitment to
specifically Euclidean-Newtonian science.
Mathematical-physical science, for Buchdahl, is the product of
the Kantian faculties of reason and judgement rather than the
faculties of understanding and sensibility. The faculties of reason
and judgement, guided by regulative principles of unity, simplicity,
and so on, aim to construct theories, conceptualizations, and laws
that inject systematic scientific order into the domain of spatiotemporally ordered particulars which is the immediate product of
the constitutive activities of the understanding and sensibility. The
nature produced by the constitutive activities of the understanding and sensibility, a mere concatenated plurality of spatiotemporal particulars, is thereby transformed into an order of
nature governed by systematic scientific laws.22 Moreover, the
scientific systems produced by the faculties of reason and
judgement, have a constructive, or postulational character, in
that they involve concepts and principles (such as basic principles
of geometry or mechanics) that are not straightforward inductive
generalizations. Such initial conceptual stipulations instead
resemble what more recent philosophers of science have called
implicit definitions, the functional a priori, or, in the
terminology of Thomas Kuhn, subject-defining paradigms.23
Like Kuhnian paradigms, however, such postulational stipulations
need in no way be fixed for all time. On the contrary, being
products of our decision and choice, we are free to revise and
transform them as the need arises. So there is not only room, on
Buchdahls reading, for a non-Newtonian physics, but even for a
non-Euclidean geometry.
What makes this loosening of the Kantian commitment to
specifically Euclidean-Newtonian science possible is a strict and
sharp separation between the nature produced by the constitutive
activities of the understanding and sensibility and the order of
nature constructed by the merely regulative activities of reason
and judgement. Buchdahl requires, accordingly, that this
separation between science (as a body of laws) and the world of
22. Op. cit., pp. 480481.
23. Op. cit., pp. 33, 510511, 676.



commonsense objects be made complete.24 For the nature

produced by the constitutive activities of the understanding and
sensibility is constituted entirely independently of all
mathematical-physical theorizing, and thus comprises the
straightforward things of commonsense bereft of all scientificotheoretical components.25 Kants Transcendental Deduction of
the Categories, in particular, is aimed only at the spatio-temporal
ordering of commonsense objects and events, and it serves merely
to supply a spatial and temporal clamp or conceptual clamp
making this previously indeterminate ordering both determinate
and objective.26 The Transcendental Deduction in particular, and
the Transcendental Aesthetic and Analytic more generally, are
thus in no way concerned with any particular mathematicalphysical scientific theory; indeed, if Buchdahl is correct, they are
not concerned with scientific theorizing at all.
We have seen, however, that such a reading of the
Transcendental Aesthetic and Analytic is untenable. For not only
is Kants own argumentation explicitly framed by the questions
How is pure mathematics possible? and How is pure natural
science possible? respectively, but the Transcendental Deduction,
in particular, effects a conceptual clamp uniting understanding
and sensibility precisely through the pure imagination, and thus,
in turn, through a transcendental synthesis exemplified by
geometrical construction, on the one hand, and the general
doctrine of motion, on the other. In this way, the transcendental
synthesis expressing the first application of the understanding to
sensibility injects the principles of (Euclidean) geometry and
(Newtonian) mechanics into the sensible appearances, and
therefore makes possible their determinate spatio-temporal
ordering precisely in so far as they are thereby subject to
mathematical-physical scientific laws. Indeed, it is otherwise
entirely obscure what the notion of spatio-temporal clamp or
conceptual clamp could possibly mean here. Buchdahls
separation between the constitutive domain of the understanding
and sensibility and the constructive activity of scientific theorizing
leaves us with an insecure grasp on the characteristically Kantian
24. Op. cit., p. 659.
25. Op. cit., pp. 638639, note 4.
26. Op. cit., pp. 621, 635.



conception of constitutive, synthetic a priori knowledge, and an

insecure appreciation, therefore, of what Kant himself takes to be
the peculiarly transcendental function of philosophy.

Buchdahls interpretation of the constructive activity of scientific
theorizing, by contrast, is quite insightful and suggestive, for it in
fact indicates a way in which the Kantian conception of synthetic
a priori knowledge can be illuminatingly generalized beyond its
Euclidean-Newtonian context. Indeed, such a generalization of the
Kantian a priori, whereby it loses its rigidly fixed character but
retains its essential constitutive function with respect to empirical
knowledge, was actually common coin within early twentieth
century scientific philosophy.
Thus Henri Poincar, for example, developed a radically new
interpretation of the status of geometry, based on his own
fundamental work on non-Euclidean spaces, according to which
geometry is neither (pace Kant) a synthetic a priori product of our
pure intuition nor (pace Gauss and Helmholtz) a straightforward
empirical description of what we can experience in nature.
Establishing one or another system of geometry rather requires a
free choice, a convention of our own in order to bridge the
irreducible gulf between our crude and approximate sensory
experience and our precise mathematical descriptions of nature.
And Hans Reichenbach, to take a second example, accordingly
distinguished two meanings of the Kantian a priori: necessary and
unrevisable, fixed for all time, on the one hand, and constitutive
of the concept of the object of knowledge, on the other.27 He
argued, on this basis, that the great lesson of the theory of relativity
is that the former meaning must be dropped while the latter must
be retained. Relativity theory, that is, involves a priori constitutive
principles as necessary presupposition of its properly empirical
claims, just as much as did Newtonian physics, but these principles
have essentially changed in the transition from the latter theory to
the former. So what we end up with, following out both of these
suggestions, is a relativized and dynamical conception of a priori
mathematical-physical principles, which change and develop
27. H. Reichenbach, Relativittstheorie und Erkenntnis Apriori (Berlin: Springer, 1920).



along with the development of the mathematical and physical

sciences themselves, but which nevertheless retain the characteristically Kantian constitutive function of making the empirical
natural knowledge thereby structured and framed by such
principles first possible.
Within more recent philosophy of science, as Buchdahl himself
intimates, we find very clear descendants of this conception: most
notably, in Rudolf Carnaps distinction between changes of
language or linguistic framework, on the one side, and revisions
of properly empirical statements formulated within a given
linguistic framework, on the other, and in Thomas Kuhns closely
related distinction between revolutionary changes of scientific
paradigm and the problem-solving activities of normal science.
Indeed, Kuhn explicitly acknowledged the intimate kinship
between his conception and that of the relativized or conventional
a priori arising in the work of such logical empiricist philosophers
as Reichenbach and Carnap, and thus that between his conception
and Kants a priori when the latter is taken in [a] relativized
Where we cannot follow Buchdahl, I believe, is in his attempt,
in addition, to defend a non-relativized, non-dynamical version of
the Kantian constitutive a priori, produced, supposedly, by the
faculties of understanding and sensibility acting independently of
all constructive mathematical and physical theorizing, and
confined, accordingly, to the spatio-temporal structure of
commonsense as opposed to scientific knowledge. This attempt not
only fundamentally distorts the Kantian texts, as I argued above,
but it also fundamentally distorts our own relation to the Kantian
enterprise. Kant lived at a time when the most basic
presuppositions of contemporary science, Euclidean geometry and
Newtonian mechanics, could be very plausibly held to be
absolutely fixed conditions of the possibility of all experience in
general. For Euclidean geometry and Newtonian mechanics, in
Kants time, simply had no conceivable alternatives. But now that
we have lost this position of innocence, I believe, we should no
longer persist, nonetheless, in the ambition to articulate timeless
philosophical truths delimiting the spatio-temporal structure of all
28. T. Kuhn, Afterwords, in World Changes, ed. P. Horwich (Cambridge: MIT, 1993),
pp. 313315, 331332.



human experience in general. We should recognize, rather, that the

Kantian conception of the peculiarly transcendental function of
philosophy must itself be relativized and generalized.
Thus our modern conception of a dynamical, yet still
constitutive a priori is the product of our own time and our own
intellectual situation. It does not make sense in the context of
Kants eighteenth century situation. Although it is true, as
Buchdahl rightly emphasizes, that Kants conception of the
regulative use of reason is quite modern in its open-endedness and
fluidity, Kant himself explicitly distinguishes the regulative
principles governing this indefinite progress of reason (simplicity,
unity, and so on) from the constitutive principles, produced by the
interaction of understanding and sensibility, that articulate the
fixed background spatio-temporal structure of all possible
empirical science. There can be very little doubt, as I have
repeatedly urged above, that this fixed background structure
essentially includes, for Kant, the principles of Euclidean
geometry and Newtonian mechanics. And there is also very little
doubt, as I have just suggested, that this is by far the most plausible
and insightful philosophical reading of the situation in the context
of eighteenth century science.
Throughout the nineteenth century, by contrast, there were
repeated attempts by mathematicians, natural scientists, and
philosophers to break free from such constraintsboth from the
constraints of Euclidean geometry and from the corresponding
constraints of Kantian epistemology. For the leading thinkers of
this period struggled to articulate new mathematical conceptions
that would generalize and transform traditional Euclidean
geometry, and also new philosophical conceptions that would
generalize and transform the doctrines of Kant. These
developments culminated, in the early years of the twentieth
century, in Einsteins theory of relativity, wherein the radically
new geometrical ideas developed in the preceding century are
worked up into a new mathematical-physical framework for space,
time, and motion capable of competing with, and eventually
replacing, the Euclidean-Newtonian framework. And the
corresponding dynamical, explicitly relativized conception of a
priori constitutive principles developed by such philosophers as



the logical empiricists was then articulated in self-conscious

response to precisely these new scientific developments.
What we see here, then, is that the enterprise Kant called
transcendental philosophythe project of articulating and philosophically contextualizing the most basic constitutive principles
defining the fundamental spatio-temporal framework of empirical
natural scienceis just as dynamical and historically situated as
are the mathematical-physical constitutive principles which are its
object. Indeed, the relationship between transcendental philosophy
in this sense and the mathematical-physical principles whose
possibility it examines and articulates is best seen as thoroughly
dialectical. Not only do new developments in mathematicalphysical constitutive principles lead to new developments in this
philosophical enterprise (as the Newtonian synthesis led to Kant,
say, or Einstein to the philosophy of logical empiricism), but the
influence, and interaction, flows in the other direction as well.
Thus, the Newtonian synthesis would itself not have been possible
without the earlier attempts to articulate and philosophically
contextualize the spatio-temporal framework of the new
mathematical science by such thinkers as Descartes and Leibniz,
without whom, we might add, the transcendental philosophy of
Kant would also not have been possible. Similarly, Einsteins
development of the theory of relativity was deeply influenced by,
albeit partly by way of reaction against, the earlier conventionalist
philosophy of geometry articulated by Poincar, which also, of
course, greatly influenced the subsequent philosophy of logical
empiricism. And, finally, the nineteenth century developments in
mathematics, natural science, and philosophy that eventuated in
the theory of relativity (including the work of Poincar) were
themselves often explicitly motivated by attempts either to
generalize or radically transform the philosophy of Kant.
Yet this emphatically does not mean, as contemporary
philosophical naturalism would have it, that philosophy simply
becomes absorbed into empirical natural science as one more
component of an holistically conceived web of belief. On the
contrary, just as our most sophisticated contemporary
historiography of science demands that we maintain a distinction,
in Kuhns terminology, between change of paradigm and normal
science (and thus, in Kantian terminology, between constitutive



principles and properly empirical claims framed against the

background of such principles), so our present picture of the
peculiarly transcendental function of philosophy demands that we
maintain a further and parallel distinction between philosophical
changes and properly scientific ones. If revolutionary science,
wherein we change the very paradigm or framework within which
the problem-solving activities of normal science proceed, takes
place at one level removed, as it were, from these problem-solving
activities themselves, then transcendental philosophy in the
present sense, wherein we articulate and philosophically
contextualize the paradigms or constitutive frameworks developed
in revolutionary science (as Kant articulated and contextualized
the Newtonian framework, say, or the logical empiricists did the
same for the Einsteinian), takes place rather at two levels removed.
In this way, despite being just as dynamical and historically
situated as are the scientific constitutive principles which are its
object, transcendental philosophy in this sense retains a distinctive
and absolutely essential role in the ongoing dialectic of
knowledge. Perhaps that is distinction enough.