1 Introduction
Immanuel Kant’s Critique of Pure Reason contains many
sections and concepts that still drive lively contemporary debate. One
of the most active of these concerns Kant’s position on idealism.
In modern philosophy, idealism is commonly understood to mean either
that the ultimate foundation of reality is the mind and nothing exists
beyond the mental, or that something does exist independently of the
mind but that our knowledge of mind-independent reality is at best
inferred and at worst entirely inaccessible (Guyer and Horstmann 2022,
sec. 1). Kant’s idealism is related to these two formulations,
but, as we will see, it is very much its own thing, and the two major
interpretations of it disagree about what that thing is.
A few pieces of terminology recur throughout this essay, and the
Kantian senses of these terms often diverge from contemporary usage.
First, transcendental: knowledge “that concerns the a priori
possibility of knowledge, or its a priori employment” (Kant,
1781&1787/2011, A56/B81). The remaining terms come
from the transcendental aesthetic in the first part of the
Critique. Intuition is the seeing of some object
(Kant, 1781&1787/2011, A19), made possible by
sensibility, the capacity for sensing things in a given
modality (Kant, 1781&1787/2011, A19). Intuitions
gained from sensibility yield concepts through the
understanding; concepts are purely thought but can exist only in
relation to intuition. An appearance is a thing as given in
intuition (Kant, 1781&1787/2011, A20/B34), and we are
limited to those intuitions our sensibility makes available. Kant
identifies the most basic, a priori aspect of our sensibility,
the ordering our minds cannot help but impose on the world, as space
and time (Kant, 1781&1787/2011, A22). Crucially, on
the Kantian view, space and time are not things that exist in
themselves. As Kant puts it, “time is the formal a priori
condition of all appearances whatsoever. Space, as the pure form of all
outer intuition” (Kant, 1781&1787/2011,
A34). The function of space and time as our a priori sensibility, what
Kant terms the transcendental aesthetic, will be central to the later
discussion.
It is also useful to introduce the two main interpretive camps on
Kant’s idealism. We will use Henry Allison’s terminology. The first camp,
Allison’s own, he calls the ‘two-aspect’ reading (Allison 2004, 3).
On this view (which later sections will develop in detail), Kant holds
that external things exist and are given in appearance, and that they
are considered in two non-ontological ways: as they appear and as they
are in themselves (Allison
2004, 12). The second camp, the ‘two-world’ reading (Allison
2004, 3), distinguishes the appearance of an object available to
our sensibility (the phenomenon) from the object itself (the noumenon),
to which we have no direct sensible access and whose existence we can
only infer. On this reading, we can know nothing about the intrinsic
properties of the noumenon, only the phenomenon. With the Kantian
terminology and the two interpretive camps in hand, we can turn to the
structure of the essay.
The first section examines how Kant views idealism, with a specific
interest in Descartes’ and Berkeley’s conceptions of it, which Kant
labels sceptical and dogmatic idealism respectively. We look at what
Kant takes the problems with these two forms to be, and then follow his
argument for refuting both at once and how that refutation fits into
the wider Kantian picture.
The second section continues with Kant’s case for his new system,
transcendental idealism, as developed in the fourth paralogism. We pay
particular attention to Kant’s use of transcendental idealism to
critique Descartes, and to what Kant sees as the problem in Descartes’
system that motivated his own position.
The third section looks more closely at the position Kant has
developed, and at the arguments offered by both sides of the
interpretive dispute over transcendental idealism. We consider what
problems transcendental idealism solves and which new problems it
raises under the two major readings. Settling that debate is well
beyond the scope of this essay, but we aim to stake out a position and
justify our interpretation against the major readings and the
Critique itself.
Finally, we turn to a modern debate in science, namely how to interpret
quantum physics. We compare two kinds of interpretation and argue that,
in the context of his idealism, Kant would have reasons to favour one
over the other.
2 Kant’s refutation of idealism
This first section examines what Kant thinks of idealism, focusing on
the two species he identifies, sceptical and dogmatic idealism. We lay
out the issues Kant takes each to suffer from and his strategy for
refuting both at once with a single proof.
Kant begins his refutation of idealism by specifying exactly which
theories he is targeting. He defines problematic idealism as
the theory proposed by Descartes (which we will call sceptical
idealism, in line with more contemporary usage), namely “the
theory which declares the existence of objects in space outside us
[...] to be merely doubtful and indemonstrable” (Kant,
1781&1787/2011, B274). On this view, we can never be
certain that the external world exists, or that it is as we perceive it
to be. Kant then distinguishes Descartes’ idealism from Berkeley’s,
which he calls dogmatic idealism. Dogmatic idealism holds
that external objects in space are “false and impossible” (Kant,
1781&1787/2011, B274); space itself is impossible,
and so any objects contingent on space are equally impossible. Kant
takes the dogmatic position to follow from the belief that space is a
property belonging to things in themselves, in which case, he says,
space is a “non-entity” (Kant, 1781&1787/2011, B274).
The key to undermining dogmatic idealism has therefore already been
laid down in the transcendental aesthetic (Kant,
1781&1787/2011, B274): for Kant, space is not within
things or a thing in itself but part of our a priori sensibility, the
fundamental way our cognition conditions our experience. The tension
between dogmatic idealism and Kant’s system is already visible.
Before launching his refutation, Kant outlines what the proof must
achieve. He must show that we have experience of external things. To
do this, he grants Descartes the primacy and infallibility of inner
experience, and then aims to show that inner experience depends on the
existence of outer experience (Kant, 1781&1787/2011,
B275). With this dependence as the criterion for success, Kant begins
his proof.
Kant claims that we are conscious of our existence, and that this
consciousness is determined through the a priori sensibility of time
(Kant, 1781&1787/2011, B275). He argues that there
must be something ‘permanent’ in perception; otherwise no determination
of time is possible. Since Kant is determining his own conscious
existence in time, the perception of the permanent is only possible
through a thing outer to oneself (Kant, 1781&1787/2011,
B275). Consciousness of our own experience in time requires the
determination of that time, and that determination is only possible
through something external that is permanent (in some manner), so that
time can be measured at all. Consciousness of our own existence
therefore requires immediate consciousness of external things (Kant,
1781&1787/2011, B276). Kant has used the
transcendental aesthetic, the a priori sensibility of time, together
with the immediate certainty of inner experience to prove the existence
of outer experience.
We can now see how Kant uses the transcendental aesthetic to refute
both sceptical and dogmatic idealism. Sceptical idealism holds the
weaker conviction that we cannot be certain of the existence of
external experience; dogmatic idealism holds the stronger one that
space does not exist and is therefore imaginary. Kant attacks both by
appealing to the a priori sensibility of time (which dogmatic idealists
do not dispute) together with the primacy of immediate inner
experience, which both forms of idealism accept.
Through his Copernican turn, Kant has shown that space and time
condition experience and that our a priori sensibility is essential to
knowledge (Braver 2007,
36–37). By reformulating the subject-object relationship, he
produces a system in which the refutation of classical idealism falls
out as a consequence.
The next section turns to the fourth paralogism and Kant’s critique of
Descartes, and begins to flesh out transcendental idealism along with
Kant’s account of its justification and implications.
3 The fourth paralogism
In the fourth paralogism, Kant critiques sceptical idealism, and
Descartes in particular, more thoroughly, using the problems he
identifies to motivate transcendental idealism. As throughout the
Critique, Kant aims to put natural science on a justifiable
basis and expose the underlying assumptions that, on his telling, had
produced an incoherent pre-critical metaphysics.
Descartes is initially praised in the fourth paralogism: Kant agrees
that Descartes was right to limit all perception to immediate
perceptions of the inner sense (Kant, 1781&1787/2011,
A368). The inference the sceptical idealist then makes, however, is
too much for Kant, especially given the refutation of idealism just
discussed. Determination of the inner sense, Kant holds, requires the
existence of perceptions in the outer sense. Descartes has therefore
identified something true, that all perceptions are indeed in the
inner sense, but has drawn the troubling conclusion that the outer
world has a merely doubtful existence (Kant,
1781&1787/2011, A366-A367). Building on the
refutation of idealism in section 2, this
section shows how Kant broadens his system and draws out its
implications for metaphysics and knowledge.
Following the fourth paralogism, we can fix Kant’s system,
transcendental idealism, in contrast with Descartes’ sceptical idealism,
which Kant labels empirical idealism (Kant,
1781&1787/2011, A369). Transcendental idealism
regards the appearances of objects as the whole ‘thing’ of the object.
There is no thing in itself beyond representations, and space and time
are only sensible forms of our intuition (Kant,
1781&1787/2011, A369). This is opposed to
transcendental realism, the doctrine that space, time, and the objects
that exist in space and time are something in and of themselves,
independent of our sensibility (Kant, 1781&1787/2011,
A369). Descartes and almost all philosophers prior to Kant are
transcendental realists (Braver 2007, 33–34), and
Kant argues that transcendental realism is precisely what opens the
door for empirical idealism. A transcendental realist, holding that
there are mind-independent objects that are things in themselves,
concludes that, given the limited sensory modalities of humans, our
representations are wholly inadequate to establish the reality of
noumenal objects (Kant, 1781&1787/2011, A370). Belief
in an inaccessible noumenal world fuels empirical idealism, and
transcendental realism ends in a place that, for Kant, leaves natural
science without reliable footing. How can we reliably do natural
science if we are cut off from an underlying world of things in
themselves?
This concern with the rational basis of the natural sciences is a
recurring theme of the Critique (Kant,
1781&1787/2011, BXXII-BXIII). In the fourth
paralogism, Kant returns repeatedly to the trouble with Descartes’
sceptical idealism. Descartes is a transcendental realist, and Kant
sees this as putting natural science in a position we do not want to
occupy. The sceptical idealist says that we can rely only on inner
perception; outer perceptions are inferred and can therefore have only
a doubtful existence (Kant, 1781&1787/2011, A368).
The uncertainty built into the sceptical idealist’s inferences is
intolerable to Kant. He instead wants to advance the system that, as
shown in section 2,
refutes idealism and establishes a “possible certainty” regarding
outer objects and perceptions (Kant, 1781&1787/2011,
A367).
For Kant, the solution is to be a transcendental idealist. If one
grants that there is nothing beyond our representations, Kant argues,
one can also be an empirical realist, and natural science can rest on
the solid footing he wants for it (Kant,
1781&1787/2011, BXXII-BXXIII). As a transcendental
idealist and empirical realist, Kant argues that we can “admit the
existence of matter” (Kant, 1781&1787/2011, A370), via
the refutation of idealism: inner experience serves as proof of the
existence of outer experience, as shown in section 2.
In the fourth paralogism, Kant is concerned with knowledge rather than
reality. The limits of human knowledge carry more weight for him than
metaphysics, and transcendental idealism reads, on our view, more like
a project of securing a valid and justified basis for human knowledge
than anything else. We take Kant to say as much later in the fourth
paralogism: “from perceptions knowledge of objects can be generated”
(Kant, 1781&1787/2011, A376). Here Kant describes our
actual interaction with the world; we generate knowledge of outer
objects through perceptions, which, by the refutation, must exist if
inner experience does. Transcendental idealism lets us be empirical
realists. It acknowledges the limits on human knowledge imposed by our
a priori sensibility of space and time, and how this conditions
experience. We generate meaningful knowledge about external objects,
conditioned by space and time, by treating as real those perceptions
that are bound up with empirical laws. That is how science actually
operates: as an empirical realist endeavour, not the empirical idealist
one transcendental realism would force on it.
There is further support for this reading in the fourth paralogism.
From around A373 onwards, Kant aims to clear up some possible
ambiguities in language. He restates what space and time are: a priori
representations that constitute our forms of sensible intuition (Kant,
1781&1787/2011, A373). They exist only in us (Kant,
1781&1787/2011, A373), and so Kant is only willing to
say things about ‘things’ that exist for us, that is, through
perceptions that come from and are conditioned by our sensible
intuitions. He distinguishes objects one might call transcendentally
external from things found in space (Kant,
1781&1787/2011, A373), the latter being within our
sensible intuitions. Transcendental idealism, better understood
through its empirical realist face, is concerned only with what exists
in our sensible intuitions. This is further evidence that Kant has
more to say about knowledge than about metaphysics. What we can know
about reality is, for all intents and purposes, reality for us. To
try to say more is the antithesis of the whole project of the
Critique, which aims to show which areas of knowledge are
inaccessible by their very nature (Kant, 1781&1787/2011,
BXXV).
This reading of transcendental idealism as an epistemic, or
metaepistemic, system rather than a metaphysical one will return in
the next section, when we turn to the ongoing debate over
interpretations of transcendental idealism.
4 Two-aspect or two-world
In the Western tradition, Kant’s transcendental idealism is a
genuinely new position, notably the first anti-realist one (Braver 2007,
34). As we have seen, most philosophers of the sceptical idealist
stripe are transcendental realists, and dogmatic idealists deny all
external experience, which at the very least implies an underlying
inaccessibility of any reality outside the internal sense. Transcendental
idealism may seem an odd position: to refuse to regard things in
themselves at all, and to regard outer things as existing only when
they are within our sensible intuition. We turn now to Kant’s wider
treatment of noumena, since interpretations of transcendental idealism
have generally hinged on how Kant is taken to understand noumena and
their metaphysical and epistemic place in his system.
Kant conceives the noumenon in two ways, a positive and a negative
conception. He finds the former contradictory and therefore of little
interest, while the latter is conceivable but irrelevant once reality
is viewed through the lens of empirical realism. The positive
conception defines a noumenon as an “object of non-sensible intuition”
(Kant, 1781&1787/2011, B307). Kant immediately sets
this aside, since it would require a special mode of intuition that
humans plainly do not possess; we cannot conceive further forms of
sensible intuition (Kant, 1781&1787/2011, B307), and
no human possesses the ‘intellectual’ intuition the positive conception
requires (Kant, 1781&1787/2011, B307).
The negative conception holds that a noumenon is a thing that is “not
an object of our sensible intuition” (Kant, 1781&1787/2011,
B307). For Kant, this is the only viable way to conceive a noumenon
(Kant, 1781&1787/2011, A252). He grants that a
non-sensible form of intuition is conceivable, but notes that being
metaphysically admissible is not the same as having any bearing on
reality. He takes himself to have proved and described our sensible
intuition, and to have shown that nothing else is open to us. This is
all coherent within transcendental idealism, since we are only
concerned with the empirical realism of outer things within our a
priori sensibility, chiefly space in this case.
We can now look more carefully at the contemporary dispute between the
‘two-aspect’ and ‘two-world’ interpretations introduced earlier. The
‘two-world’ interpretation reads Kant as drawing a substantive
metaphysical distinction between phenomena and noumena. Henry Allison
argues that this position grows out of Kant’s subjectivist starting
point (Allison 2004, 5).
From that starting point, Kant must maintain either that things only
seem spatial or that appearances are spatial (Allison 2004, 5). Both
options are uncomfortable: the first, that appearances only seem
spatial, slides back into sceptical idealism, while the second forces
us to regard appearances (mental constructs) as extended in space (Allison 2004,
6). Allison takes the ‘two-world’ interpretation to hinge on
Kant’s wish to maintain that we only know things as they appear to us,
i.e., as phenomena. To make this compatible with empirical realism,
he shifted to saying that we know appearances and that those
appearances are spatial (Allison 2004, 6). The core
issue for the ‘two-world’ camp, Allison argues, is the meaning of
‘knowledge’ (Allison
2004, 6): on this reading, knowing something means knowing how
it is in itself, i.e., as a noumenon. But for Kant we cannot know
anything about the noumenal world, and so idealism returns with full
force, in either a sceptical or a dogmatic guise. The sceptical version
holds that we cannot know anything with certainty about the noumenal;
the dogmatic version, that since noumena fall outside our a priori
sensibility of space and time, we are certain they are unknowable
(Allison 2004, 6–7).
Consider first Allison’s response to the dogmatic two-world
interpretation, and then the sceptical one. In the dogmatic case,
proponents such as Paul Guyer argue that no extra noumenal objects are
required. We experience ordinary objects, which are to be considered
as things in themselves once our sensibility is removed, i.e., ordinary
objects with space and time stripped out (Wood et al. 2007, 14).
Things in themselves, on this view, are nothing like our representations
of them, since we conform those representations to our cognition
through space and time. Allison argues this is incoherent: it requires
reading Kant as identifying ordinary objects such as tables, chairs,
and books with things in themselves while simultaneously denying that
they have the properties we experience them to have. We find this
incoherence troubling, and the epistemic reading we have developed in
working through the text holds up better. Guyer’s reading is
uncharitable enough that it would also discard the benefits to
empiricism and natural science that Kant has built up.
Rae Langton offers the most useful version of the sceptical ‘two-world’
interpretation. On her reading, an object’s noumenal properties are
intrinsic, and we can only perceive relational properties (Langton 1998,
81). The relational properties are the phenomena, since
something must affect our sensibility to generate representations in
the inner sense; the intrinsic, noumenal properties remain unknown
because we cannot intuit them. Allison agrees that we cannot determine
the intrinsic properties of an object, but argues that the metaphysical
use of phenomena and noumena does not fit Kant’s actual concern with
the conditions for the representation of objects, which Allison calls
epistemic conditions (Allison 2004, 11). Allison’s
argument lines up with our own reading of Kant’s phenomena and noumena:
things can lie outside our conditions for intuition and experience
without being objects for us (Allison 2004, 12). Phenomena
and noumena are not distinct objects, which would make the conditions
of experience ontological (Allison 2004, 12). They are
two aspects, two different ways of considering the same things (as
they appear and as they are in themselves), rather than two
ontologically distinct categories (Allison 2004, 16).
We find ourselves most in agreement with the two-aspect view, for a few
reasons. It is the most epistemological interpretation of
transcendental idealism, and so the most consistent with what we take
to be Kant’s aim of setting up a coherent system of scientific
knowledge. It also avoids the sceptical problem of inference that
Kant himself diagnoses in the fourth paralogism (Kant,
1781&1787/2011, A368), and which the two-world
interpretation does inherit. Finally, the incoherence Allison finds
in the dogmatic two-world reading further reinforces the epistemic
approach.
With the interpretation of transcendental idealism now in place, the
final section turns to interpretations of quantum physics and asks
which one Kant would have been most likely to endorse.
5 Kant and Quantum
Some two centuries later, Kant’s transcendental idealism may be
applicable, and perhaps even partly vindicated, in quantum physics. A
persistent issue for quantum physics is metaphysical interpretation:
working out what the uncertain, probabilistic clouds that are the
wavefunctions of quantum systems imply for reality is an ongoing
debate, and the ‘collapse’ of those wavefunctions into observable
outcomes is a decades-old conceptual problem (Maudlin 2019, 97–98). The
question of what a wavefunction means is bound up with, and compounded
by, the role of the observer. In the double-slit experiment, an
electron beam is passed through two slits. When neither slit is
monitored, the electrons display wave-like behaviour and form
interference patterns on a screen behind the slits. When either slit
is monitored, i.e., the behaviour of the observer is changed, the
behaviour of the electrons also changes: they now act as particles,
pass through one slit only, and produce no interference pattern
(Maudlin 2019, 15–17). Since
our physical grasp of the wavefunction itself is fairly minimal, all
we can do is posit which behaviours of the physical observables are
most likely on measurement. There is already a hint of empirical
realism and transcendental idealism here. Within quantum theory the
observer has taken a central place in physics, and that should look
attractive from a Kantian point of view. We can be empirical realists
about the outcomes of collapsed wavefunctions, the physical
observables, while the knowledge we have of pre-observation empirical
quantities and qualities, before any subject-object interaction, is
purely statistical.
We will argue that the interpretation of quantum physics most natural
to the Kantian is Quantum Bayesianism, usually shortened to QBism.
The core claim of QBism is that the probabilities of quantum physics
(the wavefunctions) are always subjective, and so quantum states (the
physical features of the system) are themselves subjective (Timpson 2013, 188–89). Quantum
states are therefore not objective features of reality but subjective
degrees of belief, in line with the standard way of interpreting any
Bayesian programme (Timpson 2013, 189). This
formulation dissolves the measurement problem and gives a clean account
of how the observer figures in the subject-object relationship of the
quantum world. QBism is an epistemic theory: like transcendental
idealism, it commits us only to what we know, or to our credences
about the likelihood of a given outcome, as primary. The thing in
itself, the wavefunction, is meaningless for us beyond the statistical
likelihood of something we can know, namely the empirical result. If
this is not quite a modern scientific vindication of transcendental
idealism, it is at least an epistemic theory of scientific knowledge
that places the subject, conditioning her experience of the object,
at the centre. The resemblance is striking.
This epistemic theory of quantum physics is far more congenial to
Kant than the more metaphysical alternatives. The famous many-worlds
interpretation, for instance, holds that all quantum states are
equally real and exist in parallel universes with separate histories
(Maudlin 2019, 174–77), a
robustly metaphysical picture of reality. We think Kant would find
this hard to reconcile with his conception of natural science. It
occupies a transcendental realist position that, as we saw in his
critique of Descartes, is always vulnerable to scepticism, making
metaphysically realist interpretations of quantum physics a
“speculative science” (Kant, 1781&1787/2011, BXIV).
Many-worlds and other transcendental realist interpretations suffer
from being unprovable, and would, in our view, lead Kant to treat them
as akin to attempts at knowledge of God: a mere ‘mock combat’ (Kant,
1781&1787/2011, BXV).
Pushing metaphysical interpretations of quantum physics, when the
wavefunction has proven to be all the information we can gain, and
apparently always will be, suggests there is no further ‘more’ to be
found by metaphysical means. To Kant, we think the project would look
like a “groping among mere concepts” (Kant,
1781&1787/2011, BXV), especially when the QBist
epistemic approach already dissolves problems that transcendentally
realist theories continue to struggle with (Timpson
2013, 203–4).
6 Conclusions
To conclude, by combining his proof against empirical idealism with
his critique of Descartes, Kant arrives at transcendental idealism, a
system aimed at providing a solid basis for natural science. Being a
transcendental idealist is a radical move, but it dissolves the
problem of having to infer reality and the existence of outer objects
by grounding inner experience in the existence of outer experience.
This underwrites empirical realism: what we experience is true and
real, while anything beyond it is outside our knowledge and does not
exist for us.
Our reading of Kant on idealism aligns most closely with the
epistemological interpretation set out by Henry Allison, the
two-aspect view, which denies any ontological distinction between
phenomena and noumena.
Finally, we looked at the metaphysics of quantum theory and found that
contemporary QBism has several features we think would make it
attractive to Kant. It is an epistemic theory that dissolves several
of the problems faced by ontic quantum interpretations. QBism limits
what we know about quantum states and reality, and takes subjectivity
and the observer’s role in a quantum system seriously, while still
holding to empirical realism about observable outcomes. All of these
features are strikingly familiar from within Kant’s system.