Realism in the Refutation of Idealism

Andrew Brook


In the Refutation of Idealism and in a long footnote on the same subject added to the second-edition Preface, Kant seems to say things that point, prima facie, strongly in the direction of realism. Because any such view would seem to be completely incompatible with the doctrine of the unknowability of things as they are and some of his other views, few commentators have been willing to take them at face value. In this paper, we examine these indications of realism, and then propose a way to render them compatible with things in themselves being unknowable. The key move is to distinguish between being aware of something and having knowledge of it. Kant made this distinction a centrepiece of his treatment of awareness of self. Did it also enter his thinking about awareness of objects?

Kant's dominant view of the sensible foundation of knowledge is that we are immediately aware of nothing but our own representations. However, as Paul Guyer has so richly documented, a streak of direct realism can also be found in his work from time to time, a streak that would seem to be in considerable tension with the official view. In the first Critique, this streak of realism shows up most clearly in the Refutation of Idealism: he tells us at one point that we must have "an immediate awareness of the existence of other things outside me" (B276), of "an external thing distinct from all my representations" (Bxli), being careful in these statements to include both the empirical sense of externality, being located in space (`outside me', `external thing') and the transcendental sense (`other things', i.e. things other than myself, which are `distinct from all my representations').

In the first Critique the Refutation of Idealism is given in two parts. In addition to the section so named, it is taken up in a long footnote appended to the new Preface. There Kant tells us that he was not happy with some of the details of the official argument and asks that certain passages in the footnote be substituted. I will treat the original argument and the long supplementary footnote together.

The central argument of the Refutation runs as follows.(1) First, "I am aware of my own existence as determined in time" (B275). What he means by "determined in time" is unclear in the Refutation, but gets clarified in the footnote. He means the application of the apparatus of location in time to myself in any way whatsoever: recognizing earlier and later stages of myself and combining them, comparing the time of events in me to the time of other events, locating myself in time, and so on.

Secondly, I do not determine myself in time on the basis of anything represented to me about myself. When I am aware of myself as subject of experience, determinations of time are not represented at all. This form of self-awareness is

a merely intellectual representation of the spontaneity of the thinking subject. This `I' has not, therefore the least predicate of intuition, which as permanent, might serve as correlate for the determination of time in inner sense -- in the manner in which, for instance, impenetrability serves in our empirical intuition of matter [B278].

Thus, if I am going to determine my own existence in time, I could only do it via the contents of inner sense. In any case, my temporal apparatus can be applied at all only to intuitions, only to something that has a manifold, a multiplicity of items (Bxl). For me to be able to apply temporal predicates to myself, therefore, I must do so via applying it to intuitions. For this, however, not just any old intuitions will do; mere multiplicity is not enough. To apply temporal predicates, we must also be able to identify change. To identify change, however, we must be able to identify something as persisting through the change -- we must be able to identify something permanent. For this, awareness of the contents of inner sense can serve no better than awareness of self as subject.

Moreover, and this is a third and key move, by themselves and cut off from things other than ourselves (Bxxxix fn.), neither representations nor any contents of a representation could do any better at representing permanence.

... the representation of [the permanent] may be very transitory and variable like all our other representations, not excepting those of matter, it yet refers to something permanent. The latter must therefore be an external thing distinct from all my representations ... [Bxli; my emphasis].

Our representations are constantly changing; indeed, they cease altogether for a number of hours each night. Therefore, the representation of permanence cannot consist in anything permanent in representations. Instead, from the contents of various representations we must somehow extract something that we can treat as a representation of a persisting object. If this object were merely a property of myself, however, it would have no permanence either. Therefore, an object could be represented as permanent only if it is "an external thing distinct from all my representations" (Bxli); I must be aware of at least some thing that is neither a representation nor myself. "In other words, the awareness of my existence is at the same time an immediate awareness of the existence of other things outside me" (B276). At least some of the intentional objects of my representations must tell me of the existence of real, independently-existing objects. QED. Kant is now advocating some form of direct realism.(2)

Is there anything to the argument of the Refutation? It is hard to tell. Even if we grant that objects of representations have no permanence, why are they not able to represent permanence unless they represent something other in the transcendental sense than oneself? Kant says nothing to help us. Perhaps he is confusing objects of representation containing no permanence, in the sense of not being permanent, with them not being able to represent permanence. Whatever, for the argument of the Refutation, Kant must show that representations cannot represent permanence by themselves. There are other controversial premises, too, but here I do not intend to examine Kant's argument. Instead, I want to focus on the realist conclusion. What are its implications? Can it be squared with other things in the critical philosophy, in particular the doctrine of the unknowability of things in themselves?

For Kant did not give one inch on the unknowability of the noumenal in the second edition. Nor, for that matter, does he ever say that he is abandoning the idea that we are aware only of our own representations. So what are we to make of the new realism? Can having immediate awareness of "an external thing distinct from all my representations" be squared with the rest of the critical philosophy?

To begin our search, notice first that the argument of the Refutation is by no means unanticipated in the first edition, though many seem to believe the opposite. Only the location, some details of the structure, and of course the conclusion are new. When Kant turns to the Paralogisms as a whole in the first edition, immediately after the discussion of the fourth Paralogism, he says:

... the appearance to outer sense has something fixed or abiding which supplies a substratum to its transitory determinations ..., whereas time, which is the sole form of our inner intuition, has nothing abiding and therefore yields knowledge only of ... change ..., not of any object that can be thereby determined. For in what we entitle `soul' everything is in continual flux and there is nothing abiding except ... the `I', which ... has no content, and therefore no manifold ... [A381].

Kant's argument for the first Analogy, the Principle of Permanence of Substance, is likewise similar in structure to the argument of the Refutation. The same is true of the argument-structure of A108. Like the Refutation, all these passages start from self-awareness, though the Refutation starts from empirical self-awareness of myself as determined in time, not transcendental awareness of myself as myself, a point Allison makes.(3) Likewise, the fundamental idea in all these passages is that I could appear to myself as I do only if my representations have a certain character; in the case of the Refutation, "awareness of my existence is bound up by way of identity (identisch verbunden) with the awareness of ... something outside me" (Bxl).(4) Of course, the Refutation reaches a stronger conclusion than the first-edition passages. It argues that representations must represent objects external in the transcendental sense, i.e., object genuinely other than myself, whereas the first-edition passages argue only that objects must be located in space and time and tied together under the Categories. Nevertheless, at least the argument-structure of the Refutation is not a radical departure from the first edition.(5)

So what are the implications of the new doctrine? Kant's new doctrine can be split into two: as well as the new notion that we are aware of objects other than ourselves, there is a new concept of what a genuinely external object is like. Unlike the discussion of the fourth Paralogism, Kant is now drawing a deep distinction between representation of an object and at least some objects; now at least some objects are quite distinct from our representations of them. In the first edition, the distinction between `real objects independent of our representations' and `intentional objects whose existence depends on our representations' depended merely on our passivity to the former and denseness of causal integration. Now it takes on some real strength.

With this change seems to go a change in Kant's conception of matter. In the first edition, Kant treated matter as a mere feature of appearances -- a feature that consists of the objects of these appearances having extension, impenetrability, cohesion, and motion (A358) -- and contrasted it with things as they actually are (A268=B324).

Matter is with [the transcendental idealist], therefore, only a species of representations (intuition), which are called external, not as standing in relation to objects in themselves external, but because they relate perceptions to the space in which all things are external to one another, while yet the space itself is in us [A370]

What the `substrate' (A350) of matter might be like, what "inwardly belongs to it" (A277=B333, a nice Leibnizian term), is hidden from us. All we can be aware of are its effects on our representations. In the Refutation, this doctrine of matter undergoes a transformation. Having argued that we must have immediate awareness of something other than ourselves that is permanent, Kant says in Note 2. that "... we have nothing permanent ... save only matter" (B278, his emphasis). He then gives the earth and the sun as his example -- we can see the sun move by comparing it to the earth's permanence. To our immense frustration, this elusive hint is all Kant gives us, but it is enough to indicate that he now seems to believe that matter exists independently of us.(6)

Must Kant also abandon or modify his doctrine of the ideality of space? This is the doctrine that space has no extra-mental existence. Though it might still be us who impose spatial matrices, it would surely be utterly unmotivated now to continue to insist that things as they are could not have spatial properties. If so, the treasured distinction of the first edition between being external to me in space (a state compatible with being a property of me) and being an object other than me should disappear, too. Unfortunately, Kant gives us nothing to allow us to pursue these questions further, not in the first Critique at least.

So let us turn to the final question I will consider: Can the new view be squared with the doctrine of the unknowability of things as they are? One way to solve the problem would be to construe the new claims about awareness of `other things outside me' as falling within transcendental idealism. This would immediately solve the problem, and is the approach Allison takes: he construes the new awareness as merely a new application of the general doctrine of Kant's mentioned earlier, that we are aware of only representations (hereafter OR, for `only representations').(7) Guyer takes Kant's realist pronouncements more seriously, quoting his saying that we have an "intellectual intuition" of "other things outside me" which is "not a mere representation of them in space" (i.e. not intuitional). Despite this, Guyer cannot bring himself to suggest that Kant could contradict OR any more than Allison. In Guyer's view, Kant is merely claiming that we must presuppose "that there are external objects", not that we must be immediately aware of them; our representations do not actually present objects other than oneself, they just presuppose such objects.(8) So let us ask: Why does even a commentator as sensitive to the realist strain in Kant as Guyer refuse to accept his realist pronouncements at face value? What makes him foist such a complicated and implausible account on Kant?

I do not think that it could be merely because the new pronouncements are inconsistent with OR. OR is not only extraordinarily implausible, it has caused no end of mischief in the history of philosophy. Any reason to think that Kant edged away from it at some points in his career would be a reason to rejoice. Rather, I think the reason has to be that the new doctrine seems to be so blatantly inconsistent with the doctrine of the unknowability of the noumenal. Our task is to see if that is so.

Though it has been little remarked upon in the literature, Kant made a distinction between being aware of something and having knowledge of it that is vital to the question before us. Most of the time the distinction arose in connection with awareness of self of a certain kind, so let us first explore it in that context. In the first edition, he says that we can denote the self "without noting in it any quality whatsoever" (A355). In the second edition, he speaks of an "awareness of self" that is "very far from being a knowledge of the self" (B158), and that we are aware of ourselves "not as we appear, or as we are, but only that we are" (B157). Kant seems to be invoking exactly the same non-knowledge but still immediate awareness of the self in the long footnote: "I am aware of my existence in time ... , and this is more than to be aware merely of my representations" (Bxl, my emphasis). Now entertain an interesting if necessarily speculative idea: suppose Kant applied the same analysis to awareness of things other than the self? Suppose he distinguished immediate awareness of objects other than oneself from knowledge of them, too? If so, he could have his new claims about our immediate awareness of them without violating his old view that we have no knowledge of them. There is a bit of evidence to support this speculation, though not much -- Kant makes a few statements that point to it.

In the long footnote, Kant puts his new idea in a surprisingly large number of different ways. Sometimes he puts it in exactly the way we have been examining: "the determination of my existence in time is possible only through the existence of actual things which I perceive outside me" (B275-6). Sometimes he puts it in a way that does not actually imply direct realism at all: we must have "awareness of a relation to something outside me" (Bxl). But sometimes he puts it this way: we must have merely immediate awareness of "the existence of other things outside me" (B276, my emphases in all cases). This claim could easily have behind it the distinction between being aware of something and knowing anything about it that we have just explored in connection with awareness of self.

As an exception to any two-world picture of phenomena and noumena, this new view would be drastic; it would be a death sentence for OR. If we are immediately aware of the world as it is, the idea that the world as it is never appears in any way in our representations has to go. Neither implication seems to me to be fatal for a suggestion that Kant might have held, or at least have been working his way toward, the new view.

In fact, in one respect, the Refutation may go further with immediate awareness of things as they are than even the other second edition passages just cited did. In the Refutation we are not just aware of objects other than ourselves, we even have one piece of knowledge of them: that they are permanent, some of them anyway. This would mean that on this one point, our representations of the world would actually represent the world as it is. Walker has expressed a fear that allowing immediate awareness of the self would open a flood-gate to knowledge of the noumenal. So far as awareness of self is concerned, I think his worry is groundless.(9) With respect to the statements in the Refutation and the long footnote we have been examining, however, he may well have a point. Even here, Kant could still cogently insist, we have no immediate, unconstructed awareness of any other property of anything, so have no other knowledge of their properties.

Is there any reason to think that Kant might have applied his notion of a kind of `transcendental' reference to self in which no qualities are noted to things other than oneself? One reason is that for Kant, awareness of self and awareness of things other than self are symmetrical. If so, and if there is a form of reference to self that requires no description or concept-application, then Kant could well have made use of a notion of a similar form of reference to objects. On the reading of the Refutation that I am suggesting, reference to self and reference to objects other than the self would display just this symmetry. In both cases, we may have no knowledge of the things to which we refer, knowledge of them as they are, but in both cases our acts of reference would refer to and thus make us aware of the objects themselves, not just representations of them. Of oneself these acts would yield a `bare consciousness' (A346=B404) of the self that is "very far from being a knowledge of the self" (B158). Of things other than oneself, they would yield "an immediate awareness of the existence of other things outside me" (B276) that would be equally far from being a knowledge of them.

The distinction between being aware of something and knowing anything of it points to an important theory of reference. On this distinction, reference could `reach all the way' to its object, yet description could remain an act of constructive concept-application, even to the point of the constructor not being able to know whether it is ever accurate -- reference could reach a real object, free of potentially distorting judgment or description, and yet all possible room for description to be `theory-laden' and otherwise influenced by the cognitive apparatus of the mind doing the describing could be preserved. When Kant called a certain kind of reference transcendental designation (A355), he may even have had something like this in mind; when reference `notes no qualities', is non-ascriptive, it would be transcending the apperceptive, synthesizing activities of the mind. Once such an act of non-ascriptive reference is made, it would immediately be surrounded by an `umbra' of cognitive manipulations, of course: the undescribed object to which reference has been made would be judged, described, propositional attitudes would be taken up to it, theories could be formed about it, and so on. It would be at this stage but only at the this stage that we would enter the realm of knowledge. For one thing, knowledge requires the possibility of error -- incorrect judgment or description -- and there would be no possibility of this kind of error in an act of non-ascriptive reference.(10)

It would also be at this stage that we would enter the realm of what cannot be checked against things as they are, where we could now understand the latter to be the objects to which we have achieved reference. In fact, the possibilities for descriptive error within this theory of reference are vast, so vast that even something as basic as how I carve the world up into objects could be in error. But what would not be in error when I have achieved reference is a belief that I am referring to and therefore am aware of something -- something other than myself. This sort of theory of reference is quite different from the picture generally accepted in Anglo-American philosophy since WWII, in which reference is always under a description. However, it or a view like it does have contemporary proponents, including Putnam, Kripke, and the later Wittgenstein. It is at the heart of most paradigm-based semantics theories. If I am right, once again Kant proves to be more than a cultural artefact, a mere earlier stage in our intellectual history.