I agree with much in Lucy Allais’s thoughtful and original Manifest Reality: Kant’s Idealism and his Realism (Oxford, 2015), including her interpretation of transcendental idealism as a moderately metaphysical position, to be identified with neither a pure Berkeleian phenomenalism nor the merely methodological interpretation of Graham Bird and Henry Allison, and with her location of the fundamental argument for transcendental idealism in Kant’s Transcendental Aesthetic rather than in the Transcendental Analytic. However, her reconstruction of the argument for transcendental idealism in the Transcendental Aesthetic is incomplete, omitting Kant’s most crucial but also most problematic step, and is actually incompatible with her moderate metaphysical interpretation. Her reconstruction turns on the extension, from the case of pure to that of empirical intuitions, of the assumption that the immediate presentation of objects by intuition is a direct presentation, but this is in tension with much of what Kant says about empirical intuitions. Allais’s reconstruction also omits Kant’s attempt to forestall what was to become the most famous objection to his position – although his attempt was ultimately unsuccessful.
The phenomenalistic position is that all we can know is our own representations, that things in themselves would be objects ontologically distinct from our representations of them, or mind-independent, thus that we cannot know anything about them, not even that they exist, and thus that Kant’s undeniable posit of things in themselves is unjustified and gratuitous. The deflationary position is that there is no ontological difference between appearances and things in themselves at all, rather that we merely have two different conceptions of things: the conception of them as appearances, that is, a conception that includes the necessary features of the ways in which things appear to us, namely spatiality and temporality; and a conception of things as things in themselves, which omits those ‘epistemic conditions’ (as Allison called them) precisely in order to drive home that they are epistemic conditions. But merely omitting properties from a concept of things does not change anything about the things themselves, so the concept of things in themselves does not strip things themselves of any of the properties that they appear to have, any more than excluding information about gender from a job application actually neuters anyone; thus the distinction between two conceptions does not entail any ontological distinction between appearances and things in themselves. And Kant could hardly have intended to derive the non-spatio-temporality of things in themselves from their concept, since that would yield only an analytical proposition. I mention this last point because I think that Allais’s omission of Kant’s final step in his argument for transcendental idealism exposes her reconstruction to a similar risk.
The moderate metaphysical position, by contrast to both of these, holds that there are things that exist independently of our representations of them, that these things necessarily appear to us in certain ways (spatially and temporally), but that these things do not themselves have the properties they appear to us to have (spatiality and temporality), yet that they must have some properties, or there must be ‘a way they are’, and even that the latter must in some way ground the former, that is, the properties they have must ground the way they appear to us to be, even though we cannot know what these properties are and thus cannot know how they do this. This position is metaphysical insofar as it asserts the existence of mind-independent things in addition to our mind-dependent representations of things, but is doubly moderate, first insofar as it does not, radically, deny the existence of mind-independent things, and second insofar as it does not, radically, hold that we can have knowledge of the properties of such things by means of pure (theoretical) reason as contrasted to intuition. It is the position that we can, indeed must, assert the existence of things that do not depend on our representations even though we can only determinately attribute to them properties that do depend on our representations, since all the properties that we attribute to things are spatial and/or temporal in form and, Kant holds, spatiality and temporality themselves are representation-dependent properties.
Kant makes it amply clear that this is how he understands transcendental idealism, especially in the Prolegomena to Any Future Metaphysics (see §13, note II, 4: 289, and note III, 4: 293). The question is, why does Kant prefer such a form of transcendental idealism to a naïve realism that would hold that we do perceive mind-independent objects and that they are much as they appear to us to be?
A basic claim of Allais’s moderate metaphysical approach is that the existence of things in themselves is not a theoretical claim that needs justification, but a natural assumption that enjoys a presumption of innocence. Kant does not take thinking that there are things in themselves ‘to be a technical or unusual notion which requires introduction and definition: he simply starts with the idea that there are things and that things have a way that they are independent of their relations to other things, including us’ (p. 34). In support of this position, she discusses Kant’s analogy between spatiality and temporality on the one hand and colour on the other, the point being that, just as removing colour from the object to the subject does not deny the existence of coloured objects, likewise removing spatiality and temporality from the object to the subject leaves the existence of the external object untouched. Just as since the seventeenth century we have all supposed that colour is a feature only of the way things appear to us that is grounded in properties of those things that do not resemble their coloured appearance, without denying the external and independent existence of the things that give us the experience of colour, so too we can suppose that spatiality and temporality are features, indeed essential features, of the way things appear to us, not properties of those things as they are in themselves, without having any reason to deny the existence of those things. It is in this discussion that Allais’s title term ‘manifest reality’ comes into play. Like colour, she holds, spatiality and temporality are ‘essentially manifest’ properties, properties only of the way things appear rather than the way they are, but that gives us no reason to deny the natural assumption that there are things that appear to us in these ways and that they have some ways they are in themselves, which ground the way they appear to us, even though these properties are not the same as the ways they appear to us and we therefore cannot understand how they ground the spatio-temporal way things appear to us. (This is a difference with the case of colour, where we do understand at least to some degree how reflectance properties of the object, which we can further understand in terms of shapes or forces, cause us to have experiences of colour.) I would find it more intuitive to call properties manifest only in appearance ‘essentially manifestational’ rather than ‘essentially manifest’, because to my ear the latter suggests that properties that things really have are manifest or evident in our experience of them, in other words, a kind of realism, while the former suggests that these properties are generated in the appearance of things to us, which is the position that Allais holds.
This may seem to be a terminological quibble, but it does raise a flag about Allais’s tendency to blur her own distinction between relational and direct accounts of perception. My suggestion that Allais should have characterized her approach as ‘essentially manifestational’ rather than ‘essentially manifest’ is motivated by the thought that the former term better brings out that spatiality and temporality are essentially dependent upon a relation between subjects and objects. While the latter term most naturally suggests that spatiality and temporality are real properties of things as they are in themselves that become manifest to us in our experience of those things, the former more clearly suggests that these properties arise from the relation of those things appearing to us, that is, that they are properties emergent from the manifestation of things. Allais does call her interpretation of transcendental idealism relational, and that does seem to be a natural way to describe a view according to which things have a way that they are in themselves, but that is not spatio-temporal, with spatio-temporality instead emerging only in those things appearing to us, in other words, in their relation to us. As she puts it, ‘A perceptual mental state is not merely a modification of the inner state of a subject but a relational state essentially involving the object and a conscious subject’ (p. 106).
Yet Allais also characterizes her interpretation as a direct theory of perception, in a way that seems incompatible with a relational or manifestational approach, although perhaps not with an essentially manifest one, and her interpretation of Kant’s argument for transcendental idealism is going to put weight on this move. She says that ‘We need to explain the possibility of directly perceiving features of objects that are mind-dependent but not in the mind, and to explain how we could directly perceive outer things while not perceiving them as they are in themselves’ (p. 103), and that on her ‘relational view of perception ... the external object perceived is a constituent of perceptual mental states’ (p. 105, emphasis added). This seems puzzling, but the puzzle is not just that Allais calls a relational view a direct view when those seem natural opposites; the puzzle is rather that while she will leave things in themselves in a relation to our representations of them, her argument for the transcendental ideality of spatio-temporality will depend on the assumption that not only the pure intuitions of space and time but also the empirical intuition of spatio-temporal things are in the sense defined direct rather than relational. This feature of Allais’s reconstruction may prove problematic, because on Kant’s account pure intuition does not present fully determinate objects at all, and thus pure intuition cannot provide a good precedent for the claim that the immediacy of empirical intuition can also be understood as containment of its objects or their properties. The construction of forms in pure intuition does not amount to determinate cognition of real objects, and thus does not ground a proof of the direct presence and thus the mind-dependence of any properties of real objects in empirical intuition. For this reason, Kant’s argument for transcendental idealism will need the final step that Allais omits.
So let us now turn to Allais’s reconstruction of this argument. It is not to be found, of course, just in the analogy with colour, for even if this can suggest how to leave the existence of things in themselves untouched, it cannot motivate stripping those things of their spatio-temporality. In Locke’s version of the primary–secondary quality distinction, colour is reduced to a secondary quality because the best available explanation of our perception of colour requires only spatio-temporal primary qualities in the mind-independent objects, thus it leaves their spatio-temporality intact. And Kant himself insists that the analogy is ‘completely inadequate’ because colours are ‘mere alterations of our subject, which can even be different in different people’, while space and time ‘necessarily belong to [the] appearance or intuition’ of objects (Critique of Pure Reason, A29/B44–5). In other words, Kant thinks that we derive the empirical subjectivity of colours from their variability and contingency, but that we derive the transcendental ideality of spatiality and temporality from their necessity: ‘Taste and colours are by no means necessary conditions under which alone objects can be objects of the senses for us’, but space and time, as ‘condition[s] of outer objects, necessarily [belong] to their appearance or intuition’ (A28–9/B44). This point is crucial, because the final step in Kant’s argument for his moderately metaphysical transcendental idealism that Allais’s reconstruction omits turns precisely on Kant’s interpretation of his claim that the fundamental propositions about the spatio-temporality of objects, including but not limited to the mathematical statements of geometry and arithmetic, are necessarily true of their objects. Without this we cannot see why Kant insists upon the mind-dependence of spatio-temporality, and thus why he can make the imperfect analogy with colour in the first place.
So now at last we turn to Allais’s reconstruction of Kant’s argument for transcendental idealism. It begins with the claim that we need intuitions as well as concepts for cognition, that is, immediate and singular presentations of objects as well as general marks under which to subsume them. Nobody has ever denied that this is a correct interpretation of Kant, not even the German idealists who thought that it is false. But Allais goes further than the normal insistence that cognition of determinate objects requires intuitions as well as concepts when she supposes that intuitions immediately present spatio-temporal objects in a sense of ‘immediately presenting’ on which, as already quoted, ‘the external object perceived is a constituent of perceptual mental states’ (p. 105). The starting-point of Allais’s reconstruction of Kant’s argument is that ‘Kant thinks that it follows from what an intuition is (the fact that it involves the presence to consciousness of what it represents) that an a priori intuition could not present a mind-independent feature of reality’ (p. 177). More fully, here is the argument she ascribes to Kant:
1) An intuition is a representation which presents its object to consciousness.
2) What a priori intuition represents is present to consciousness – because it is intuition.
3) Objects which are independent of us can be present to consciousness only through affecting us.
4) An a priori intuition does not involve anything affecting us – because it is a priori.
5) Therefore what a priori intuition presents us with is not something that exists independently of our representing it (or of our being able to represent it).
6) Our representations of space and time are a priori intuitions.
7) So our representations of the structure of space and time do not present us with a mind-independent feature of reality: they are merely the forms of our intuition and do not represent things as they are in themselves. (p. 195)
In short, the argument is that a priori intuitions are not the product of our being affected by mind-independent objects, that our representations of space and time are a priori intuitions and therefore not the product of our being affected by mind-independent objects, that for this reason space and time themselves cannot be representations of mind-independent objects or their properties, and therefore that our ‘representations of the structure of space and time’ – that is, of spatio-temporality – cannot present us with a mind-independent feature of reality, thus with a feature of any mind-independent objects. Thus not only must space and time themselves be features of our intuition only, but also the spatio-temporality of empirical objects, objects that appear to us as in space and time, also cannot be mind-independent features of mind-independent objects. The last step is crucial for Kant’s argument, because transcendental idealism is not limited to the objects of pure intuition, if there are such.
Now any defensible argument for transcendental idealism would need to block the famous ‘missing alternative’ objection of H. A. Pistorius and F. A. Trendelenburg that spatiality and temporality might be features of both our intuitions (thus knowable a priori) and of the things we intuit. The indisputable premise that a priori intuitions are not produced by the influence of or affection from external objects (Allais’s step (4), and the turning point in her proof) does not by itself block this objection, because it leaves open the possibility that space and time are mind-independent things, or that spatiality and temporality are properties of mind-independent things, even if our a priori intuitions of space and time are not produced by the influence of such mind-independent things. What needs to be excluded, first in the case of a priori intuitions and then in the case of empirical intuitions, is thus that spatio-temporality be present in any mind-independent objects even if those properties of the latter are not presented in pure intuition because of ‘influence’ or ‘affection’.
The only thing in Allais’s reconstruction that appears designed to block the missing-alternative objection is her interpretation of ‘present to consciousness’ in step (1) as the literal containment of the present object in its presentation: if the object presented by an intuition is contained in it, it cannot be present outside the intuition. But to start the argument from this premise would be to beg the question of transcendental idealism; or, more politely, the claim that intuitions contain their objects is her definition of intuition, her interpretation of what Kant means by immediacy, yet Kant’s argument for transcendental idealism cannot rest solely on a definition, in particular on this definition. For arguments from definition yield merely analytic results, and there is no hint that Kant intends transcendental idealism to be analytically true. Thus Allais’s reconstruction does not soundly prove the transcendental ideality of space and time even in the case of a priori intuition, a fortiori it cannot use that conclusion subsequently to infer the transcendental ideality of the spatio-temporality of empirical objects (step 7). Something is missing.
Further, even if the idea of the immediacy of intuition could bear the weight that Allais places on it when she interprets it as containment or the direct presentation of a property that cannot exist in any form except as its own presentation in the case of pure intuition, we will see that Kant could not have conceived of immediacy as containment in the case of empirical intuition and thus could not have meant his proof to turn solely on the equation of immediacy with containment.
The interpretation of the immediacy of intuition as the containment or presence of its object in the intuition might seem to make sense in the case of the construction in pure intuition of what we might be tempted to call mathematical objects, even of space and time themselves as mathematical objects (infinite given wholes). Even this assumption is actually problematic, however, because Kant claims that mathematics does not yield knowledge of objects at all except when it is applied to empirical intuition (B147). Moreover, this seems true. Determining that 7+5=12 in pure intuition does not determine any object, because it does not tell us 7+5 of what=12 of what. The construction of a triangle in pure intuition, even let us say of an isosceles triangle, in which it is fixed that each interior angle equals 60°, does not yield knowledge of any particular object: it gives us the form of a possible object, but it does not present us with a determinate object because it not tell us whether a side of any actual triangle is six inches or six miles long, whether it is made out of plastic or drawn with chalk on a slate, and so on. Mathematical constructions, even space and time themselves as mathematical constructions, are not particular, fully determinate objects, thus even if such constructions are contained in pure intuitions, objects are not. We thus have to be careful about transferring a containment conception of the immediacy of intuition from pure to empirical intuition, since pure intuition does not properly contain objects.
But even if the containment interpretation of immediacy were accepted in the case of pure intuition, it would be problematic to extend it to empirical intuition. And certainly in the case of empirical intuition Kant cannot have intended to derive transcendental idealism from a mere definition of intuition, because except for one retracted passage that is not even how he defines the immediacy of empirical intuition. When in the Fourth Paralogism of the first edition of the Critique of Pure Reason Kant says that ‘Every outer perception . . . immediately proves something real in space, or rather is itself the real’ (A375, emphasis added), he asserts something that sounds like Allais’s containment-interpretation of immediacy. But Kant dropped that passage from the second edition; in the Prolegomena he assumed that we have adequate grounds to acknowledge the existence of objects distinct from our representations of them, represented by our representations, although not represented by our representations in the way that they actually are; and in the Refutation of Idealism in the second edition Critique that replaced the Fourth Paralogism of the first, he insisted precisely that such a relational view of spatio-temporality is consistent with the immediacy of empirical intuition. He thus rejected a containment-interpretation of the immediacy of empirical intuition.
On my own view, Kant’s argument for transcendental idealism certainly starts with the assumption that our pure intuitions of space and time are not due to external influence, but he does not rest his argument for transcendental idealism on that fact. Rather, he concludes his argument with a final step that is aimed precisely to block the missing-alternative objection. It goes as follows. In what he labels the Metaphysical Expositions of space and time in the second edition reorganization of the Transcendental Aesthetic, Kant argues that we have an a priori intuition of both space and time themselves, as infinite given wholes (arguments 3 and 4), as well as of spatio-temporality as the form of all objects given in space and time (arguments 1 and 2). He thus assumes from the outset not merely that we can construct mathematical objects in our pure forms of intuition but also that these pure forms are the forms of our experience of empirical objects. In the Transcendental Expositions, Kant makes it clear that his conviction of our a priori knowledge of the spatio-temporal forms of empirical objects does not depend upon the assumption that we have synthetic a priori cognition in mathematics (geometry and arithmetic), although he assumes that we do have such knowledge and that our a priori intuition of the pure forms of space and time explains it. He then infers from the premise that we have such a priori knowledge of the form of space and time and of the spatio-temporal form of objects that ‘Space [or time] represents no property at all of any things in themselves nor any relation of them to each other, i.e., no determination of them that attaches to objects themselves and that would remain even if one were to abstract from all subjective conditions of intuition’, and therefore that ‘Space [or time] is nothing other than merely the form of all appearances of outer sense, i.e., the subjective condition of sensibility’ (A26/B42; cf. A32–3/B49–50). Why does he make this inference? Because, he says, ‘neither absolute nor relative determinations can be intuited prior to the existence of the things to which they pertain, thus be intuited a priori’ (A26/B42). In one sense, this is trivially true: if to intuit something means to be given it, then of course I cannot intuit something before I am given it. But this would seem to apply only to the intuition of objects. When it comes to properties, it might seem that I could intuit them before being given any particular object that has them; to take an empirical case, given that I know that human beings can hear sounds only between the frequencies of 20 Hz and 20 MHz, I can know in advance of hearing it that any sound I do hear will fall within that range.
Kant seems to need a further argument for his claim that a priori intuition implies the subjectivity or mind-dependence of the properties intuited a priori. And he repeatedly offers such a further argument, the argument to which I earlier alluded when I noted Kant’s disanalogy between the subjective ideality of colours and the transcendental ideality of space and time. The argument is that what we know a priori we know to be necessarily true, and that we can know synthetic necessary truths only about our own representations, not about things that exist independently of our representations. Kant states this argument first in the General Remarks on the Transcendental Aesthetic. Here he starts from the example of a geometrical proposition supposedly known a priori and thus as necessarily true, the proposition that an enclosed figure can be constructed only with three straight lines, but the same argument could apply more generally to not specifically geometrical propositions, such as that two simultaneously existing objects can be intuited as distinct only by being intuited as located in different regions of space, or two events not located in different regions of space can be intuited as distinct only by being intuited in successive moments of time. Here is the argument:
If there did not lie in you a faculty for intuiting a priori; if this subjective condition regarding form were not at the same time the universal a priori condition under which alone the object of this (outer) intuition is itself possible; if the object (the triangle) were something in itself without relation to your subject; then how could you say that what necessarily lies in your subjective conditions for constructing a triangle must also necessarily pertain to the triangle in itself? For you could not add to your concept (of three lines) something new (the figure) that must thereby necessarily be encountered in the object, since this is given prior to your cognition and not through it. (A48/B65–6, emphasis added)
Kant’s argument is that, while spatio-temporality in general or the specific geometrical property he describes might be necessarily true of our representations, they would not be necessarily true of objects independent of our representations; but they are necessarily true throughout their domains, so they must not be true at all of objects of independent of our representations. Thus spatio-temporality and all that it involves must be only the subjective condition of representation.
The presentation of this argument in the Critique is no fluke. Kant repeats it in the Prolegomena (§13, note I, 4: 28) and in his unpublished notes (Reflection 5637, 18: 27). The claim is always that if the synthetic propositions about space that we know a priori and therefore know to be necessarily true were to hold of mind-independent things as well as of our own representations of them, we could only know them to be contingently true of the former, and therefore not know them to be necessarily true throughout their domain after all. But we do, so they must not be true of things in themselves at all.
This is Kant’s preferred argument for transcendental idealism, and it is his preferred argument precisely because it blocks the missing-alternative objection to which Allais’s reconstruction is, as we saw, open: it denies that spatio-temporal properties can be properties of things as they really are as well as of our representations of them because even though they would be necessary properties of the latter, we could at most know them to be contingent properties of the former; but we do know that spatio-temporal properties are necessary throughout their domain. Of course, even though this is Kant’s actual argument, it is not a very good argument, because there is an obvious objection to it. This is just that the possibility that spatio-temporal properties are properties of things in themselves would prove that we do not know them to be necessarily true throughout their domains after all. When faced with a modus tollens of the form ‘p implies q, not-q, therefore not-p’ – ‘the transcendental realism of spatio-temporality implies the contingency of our knowledge of it, but our knowledge is not contingent, therefore transcendental realism is false’ – we always have two choices: we can accept not-p or simply deny not-q, that is, just accept that our knowledge of spatio-temporality is not necessarily true throughout its possible domain. Of course, Kant does not countenance this alternative conclusion.
This completes my discussion of the argument for transcendental idealism itself. I close with the suggestion of a more general problem with Allais’s interpretation, namely, that I do not see how Allais can hold that her account is moderate because it is relational and yet insist that spatio-temporal objects are contained in and only in our intuitions of them. I have already conceded that the latter may be true of mathematical constructions in pure intuition, but on Kant’s view these are not actually determinate objects, and mathematics does not give us knowledge of objects at all except insofar as it is applied to empirical intuitions and their objects. As for the latter, Kant himself insists that empirical intuitions and their objects are distinct, although the empirical intuitions, unlike concepts, are our first and most direct representations of their objects. Allais’s moderate metaphysical interpretation of transcendental idealism begins by maintaining that mind-dependent representations give us mind-independent objects, although our mind-dependently spatio-temporal representations do not give us those objects as they are in themselves, which is non-spatio-temporal. Her containment-model of (empirical) intuitions conflicts with this relational model of perception, and for that reason as well as for its own lack of support ought not to be accepted. I think that we should accept Allais’s relational model of transcendental idealism, but her reconstruction of Kant’s argument for transcendental idealism is incomplete, and when we complete it, we should not accept Kant’s argument itself!
You have
Access
I agree with much in Lucy Allais’s thoughtful and original Manifest Reality: Kant’s Idealism and his Realism (Oxford, 2015), including her interpretation of transcendental idealism as a moderately metaphysical position, to be identified with neither a pure Berkeleian phenomenalism nor the merely methodological interpretation of Graham Bird and Henry Allison, and with her location of the fundamental argument for transcendental idealism in Kant’s Transcendental Aesthetic rather than in the Transcendental Analytic. However, her reconstruction of the argument for transcendental idealism in the Transcendental Aesthetic is incomplete, omitting Kant’s most crucial but also most problematic step, and is actually incompatible with her moderate metaphysical interpretation. Her reconstruction turns on the extension, from the case of pure to that of empirical intuitions, of the assumption that the immediate presentation of objects by intuition is a direct presentation, but this is in tension with much of what Kant says about empirical intuitions. Allais’s reconstruction also omits Kant’s attempt to forestall what was to become the most famous objection to his position – although his attempt was ultimately unsuccessful.
The phenomenalistic position is that all we can know is our own representations, that things in themselves would be objects ontologically distinct from our representations of them, or mind-independent, thus that we cannot know anything about them, not even that they exist, and thus that Kant’s undeniable posit of things in themselves is unjustified and gratuitous. The deflationary position is that there is no ontological difference between appearances and things in themselves at all, rather that we merely have two different conceptions of things: the conception of them as appearances, that is, a conception that includes the necessary features of the ways in which things appear to us, namely spatiality and temporality; and a conception of things as things in themselves, which omits those ‘epistemic conditions’ (as Allison called them) precisely in order to drive home that they are epistemic conditions. But merely omitting properties from a concept of things does not change anything about the things themselves, so the concept of things in themselves does not strip things themselves of any of the properties that they appear to have, any more than excluding information about gender from a job application actually neuters anyone; thus the distinction between two conceptions does not entail any ontological distinction between appearances and things in themselves. And Kant could hardly have intended to derive the non-spatio-temporality of things in themselves from their concept, since that would yield only an analytical proposition. I mention this last point because I think that Allais’s omission of Kant’s final step in his argument for transcendental idealism exposes her reconstruction to a similar risk.
The moderate metaphysical position, by contrast to both of these, holds that there are things that exist independently of our representations of them, that these things necessarily appear to us in certain ways (spatially and temporally), but that these things do not themselves have the properties they appear to us to have (spatiality and temporality), yet that they must have some properties, or there must be ‘a way they are’, and even that the latter must in some way ground the former, that is, the properties they have must ground the way they appear to us to be, even though we cannot know what these properties are and thus cannot know how they do this. This position is metaphysical insofar as it asserts the existence of mind-independent things in addition to our mind-dependent representations of things, but is doubly moderate, first insofar as it does not, radically, deny the existence of mind-independent things, and second insofar as it does not, radically, hold that we can have knowledge of the properties of such things by means of pure (theoretical) reason as contrasted to intuition. It is the position that we can, indeed must, assert the existence of things that do not depend on our representations even though we can only determinately attribute to them properties that do depend on our representations, since all the properties that we attribute to things are spatial and/or temporal in form and, Kant holds, spatiality and temporality themselves are representation-dependent properties.
Kant makes it amply clear that this is how he understands transcendental idealism, especially in the Prolegomena to Any Future Metaphysics (see §13, note II, 4: 289, and note III, 4: 293). The question is, why does Kant prefer such a form of transcendental idealism to a naïve realism that would hold that we do perceive mind-independent objects and that they are much as they appear to us to be?
A basic claim of Allais’s moderate metaphysical approach is that the existence of things in themselves is not a theoretical claim that needs justification, but a natural assumption that enjoys a presumption of innocence. Kant does not take thinking that there are things in themselves ‘to be a technical or unusual notion which requires introduction and definition: he simply starts with the idea that there are things and that things have a way that they are independent of their relations to other things, including us’ (p. 34). In support of this position, she discusses Kant’s analogy between spatiality and temporality on the one hand and colour on the other, the point being that, just as removing colour from the object to the subject does not deny the existence of coloured objects, likewise removing spatiality and temporality from the object to the subject leaves the existence of the external object untouched. Just as since the seventeenth century we have all supposed that colour is a feature only of the way things appear to us that is grounded in properties of those things that do not resemble their coloured appearance, without denying the external and independent existence of the things that give us the experience of colour, so too we can suppose that spatiality and temporality are features, indeed essential features, of the way things appear to us, not properties of those things as they are in themselves, without having any reason to deny the existence of those things. It is in this discussion that Allais’s title term ‘manifest reality’ comes into play. Like colour, she holds, spatiality and temporality are ‘essentially manifest’ properties, properties only of the way things appear rather than the way they are, but that gives us no reason to deny the natural assumption that there are things that appear to us in these ways and that they have some ways they are in themselves, which ground the way they appear to us, even though these properties are not the same as the ways they appear to us and we therefore cannot understand how they ground the spatio-temporal way things appear to us. (This is a difference with the case of colour, where we do understand at least to some degree how reflectance properties of the object, which we can further understand in terms of shapes or forces, cause us to have experiences of colour.) I would find it more intuitive to call properties manifest only in appearance ‘essentially manifestational’ rather than ‘essentially manifest’, because to my ear the latter suggests that properties that things really have are manifest or evident in our experience of them, in other words, a kind of realism, while the former suggests that these properties are generated in the appearance of things to us, which is the position that Allais holds.
This may seem to be a terminological quibble, but it does raise a flag about Allais’s tendency to blur her own distinction between relational and direct accounts of perception. My suggestion that Allais should have characterized her approach as ‘essentially manifestational’ rather than ‘essentially manifest’ is motivated by the thought that the former term better brings out that spatiality and temporality are essentially dependent upon a relation between subjects and objects. While the latter term most naturally suggests that spatiality and temporality are real properties of things as they are in themselves that become manifest to us in our experience of those things, the former more clearly suggests that these properties arise from the relation of those things appearing to us, that is, that they are properties emergent from the manifestation of things. Allais does call her interpretation of transcendental idealism relational, and that does seem to be a natural way to describe a view according to which things have a way that they are in themselves, but that is not spatio-temporal, with spatio-temporality instead emerging only in those things appearing to us, in other words, in their relation to us. As she puts it, ‘A perceptual mental state is not merely a modification of the inner state of a subject but a relational state essentially involving the object and a conscious subject’ (p. 106).
Yet Allais also characterizes her interpretation as a direct theory of perception, in a way that seems incompatible with a relational or manifestational approach, although perhaps not with an essentially manifest one, and her interpretation of Kant’s argument for transcendental idealism is going to put weight on this move. She says that ‘We need to explain the possibility of directly perceiving features of objects that are mind-dependent but not in the mind, and to explain how we could directly perceive outer things while not perceiving them as they are in themselves’ (p. 103), and that on her ‘relational view of perception ... the external object perceived is a constituent of perceptual mental states’ (p. 105, emphasis added). This seems puzzling, but the puzzle is not just that Allais calls a relational view a direct view when those seem natural opposites; the puzzle is rather that while she will leave things in themselves in a relation to our representations of them, her argument for the transcendental ideality of spatio-temporality will depend on the assumption that not only the pure intuitions of space and time but also the empirical intuition of spatio-temporal things are in the sense defined direct rather than relational. This feature of Allais’s reconstruction may prove problematic, because on Kant’s account pure intuition does not present fully determinate objects at all, and thus pure intuition cannot provide a good precedent for the claim that the immediacy of empirical intuition can also be understood as containment of its objects or their properties. The construction of forms in pure intuition does not amount to determinate cognition of real objects, and thus does not ground a proof of the direct presence and thus the mind-dependence of any properties of real objects in empirical intuition. For this reason, Kant’s argument for transcendental idealism will need the final step that Allais omits.
So let us now turn to Allais’s reconstruction of this argument. It is not to be found, of course, just in the analogy with colour, for even if this can suggest how to leave the existence of things in themselves untouched, it cannot motivate stripping those things of their spatio-temporality. In Locke’s version of the primary–secondary quality distinction, colour is reduced to a secondary quality because the best available explanation of our perception of colour requires only spatio-temporal primary qualities in the mind-independent objects, thus it leaves their spatio-temporality intact. And Kant himself insists that the analogy is ‘completely inadequate’ because colours are ‘mere alterations of our subject, which can even be different in different people’, while space and time ‘necessarily belong to [the] appearance or intuition’ of objects (Critique of Pure Reason, A29/B44–5). In other words, Kant thinks that we derive the empirical subjectivity of colours from their variability and contingency, but that we derive the transcendental ideality of spatiality and temporality from their necessity: ‘Taste and colours are by no means necessary conditions under which alone objects can be objects of the senses for us’, but space and time, as ‘condition[s] of outer objects, necessarily [belong] to their appearance or intuition’ (A28–9/B44). This point is crucial, because the final step in Kant’s argument for his moderately metaphysical transcendental idealism that Allais’s reconstruction omits turns precisely on Kant’s interpretation of his claim that the fundamental propositions about the spatio-temporality of objects, including but not limited to the mathematical statements of geometry and arithmetic, are necessarily true of their objects. Without this we cannot see why Kant insists upon the mind-dependence of spatio-temporality, and thus why he can make the imperfect analogy with colour in the first place.
So now at last we turn to Allais’s reconstruction of Kant’s argument for transcendental idealism. It begins with the claim that we need intuitions as well as concepts for cognition, that is, immediate and singular presentations of objects as well as general marks under which to subsume them. Nobody has ever denied that this is a correct interpretation of Kant, not even the German idealists who thought that it is false. But Allais goes further than the normal insistence that cognition of determinate objects requires intuitions as well as concepts when she supposes that intuitions immediately present spatio-temporal objects in a sense of ‘immediately presenting’ on which, as already quoted, ‘the external object perceived is a constituent of perceptual mental states’ (p. 105). The starting-point of Allais’s reconstruction of Kant’s argument is that ‘Kant thinks that it follows from what an intuition is (the fact that it involves the presence to consciousness of what it represents) that an a priori intuition could not present a mind-independent feature of reality’ (p. 177). More fully, here is the argument she ascribes to Kant:
1) An intuition is a representation which presents its object to consciousness.
2) What a priori intuition represents is present to consciousness – because it is intuition.
3) Objects which are independent of us can be present to consciousness only through affecting us.
4) An a priori intuition does not involve anything affecting us – because it is a priori.
5) Therefore what a priori intuition presents us with is not something that exists independently of our representing it (or of our being able to represent it).
6) Our representations of space and time are a priori intuitions.
7) So our representations of the structure of space and time do not present us with a mind-independent feature of reality: they are merely the forms of our intuition and do not represent things as they are in themselves. (p. 195)
In short, the argument is that a priori intuitions are not the product of our being affected by mind-independent objects, that our representations of space and time are a priori intuitions and therefore not the product of our being affected by mind-independent objects, that for this reason space and time themselves cannot be representations of mind-independent objects or their properties, and therefore that our ‘representations of the structure of space and time’ – that is, of spatio-temporality – cannot present us with a mind-independent feature of reality, thus with a feature of any mind-independent objects. Thus not only must space and time themselves be features of our intuition only, but also the spatio-temporality of empirical objects, objects that appear to us as in space and time, also cannot be mind-independent features of mind-independent objects. The last step is crucial for Kant’s argument, because transcendental idealism is not limited to the objects of pure intuition, if there are such.
Now any defensible argument for transcendental idealism would need to block the famous ‘missing alternative’ objection of H. A. Pistorius and F. A. Trendelenburg that spatiality and temporality might be features of both our intuitions (thus knowable a priori) and of the things we intuit. The indisputable premise that a priori intuitions are not produced by the influence of or affection from external objects (Allais’s step (4), and the turning point in her proof) does not by itself block this objection, because it leaves open the possibility that space and time are mind-independent things, or that spatiality and temporality are properties of mind-independent things, even if our a priori intuitions of space and time are not produced by the influence of such mind-independent things. What needs to be excluded, first in the case of a priori intuitions and then in the case of empirical intuitions, is thus that spatio-temporality be present in any mind-independent objects even if those properties of the latter are not presented in pure intuition because of ‘influence’ or ‘affection’.
The only thing in Allais’s reconstruction that appears designed to block the missing-alternative objection is her interpretation of ‘present to consciousness’ in step (1) as the literal containment of the present object in its presentation: if the object presented by an intuition is contained in it, it cannot be present outside the intuition. But to start the argument from this premise would be to beg the question of transcendental idealism; or, more politely, the claim that intuitions contain their objects is her definition of intuition, her interpretation of what Kant means by immediacy, yet Kant’s argument for transcendental idealism cannot rest solely on a definition, in particular on this definition. For arguments from definition yield merely analytic results, and there is no hint that Kant intends transcendental idealism to be analytically true. Thus Allais’s reconstruction does not soundly prove the transcendental ideality of space and time even in the case of a priori intuition, a fortiori it cannot use that conclusion subsequently to infer the transcendental ideality of the spatio-temporality of empirical objects (step 7). Something is missing.
Further, even if the idea of the immediacy of intuition could bear the weight that Allais places on it when she interprets it as containment or the direct presentation of a property that cannot exist in any form except as its own presentation in the case of pure intuition, we will see that Kant could not have conceived of immediacy as containment in the case of empirical intuition and thus could not have meant his proof to turn solely on the equation of immediacy with containment.
The interpretation of the immediacy of intuition as the containment or presence of its object in the intuition might seem to make sense in the case of the construction in pure intuition of what we might be tempted to call mathematical objects, even of space and time themselves as mathematical objects (infinite given wholes). Even this assumption is actually problematic, however, because Kant claims that mathematics does not yield knowledge of objects at all except when it is applied to empirical intuition (B147). Moreover, this seems true. Determining that 7+5=12 in pure intuition does not determine any object, because it does not tell us 7+5 of what=12 of what. The construction of a triangle in pure intuition, even let us say of an isosceles triangle, in which it is fixed that each interior angle equals 60°, does not yield knowledge of any particular object: it gives us the form of a possible object, but it does not present us with a determinate object because it not tell us whether a side of any actual triangle is six inches or six miles long, whether it is made out of plastic or drawn with chalk on a slate, and so on. Mathematical constructions, even space and time themselves as mathematical constructions, are not particular, fully determinate objects, thus even if such constructions are contained in pure intuitions, objects are not. We thus have to be careful about transferring a containment conception of the immediacy of intuition from pure to empirical intuition, since pure intuition does not properly contain objects.
But even if the containment interpretation of immediacy were accepted in the case of pure intuition, it would be problematic to extend it to empirical intuition. And certainly in the case of empirical intuition Kant cannot have intended to derive transcendental idealism from a mere definition of intuition, because except for one retracted passage that is not even how he defines the immediacy of empirical intuition. When in the Fourth Paralogism of the first edition of the Critique of Pure Reason Kant says that ‘Every outer perception . . . immediately proves something real in space, or rather is itself the real’ (A375, emphasis added), he asserts something that sounds like Allais’s containment-interpretation of immediacy. But Kant dropped that passage from the second edition; in the Prolegomena he assumed that we have adequate grounds to acknowledge the existence of objects distinct from our representations of them, represented by our representations, although not represented by our representations in the way that they actually are; and in the Refutation of Idealism in the second edition Critique that replaced the Fourth Paralogism of the first, he insisted precisely that such a relational view of spatio-temporality is consistent with the immediacy of empirical intuition. He thus rejected a containment-interpretation of the immediacy of empirical intuition.
On my own view, Kant’s argument for transcendental idealism certainly starts with the assumption that our pure intuitions of space and time are not due to external influence, but he does not rest his argument for transcendental idealism on that fact. Rather, he concludes his argument with a final step that is aimed precisely to block the missing-alternative objection. It goes as follows. In what he labels the Metaphysical Expositions of space and time in the second edition reorganization of the Transcendental Aesthetic, Kant argues that we have an a priori intuition of both space and time themselves, as infinite given wholes (arguments 3 and 4), as well as of spatio-temporality as the form of all objects given in space and time (arguments 1 and 2). He thus assumes from the outset not merely that we can construct mathematical objects in our pure forms of intuition but also that these pure forms are the forms of our experience of empirical objects. In the Transcendental Expositions, Kant makes it clear that his conviction of our a priori knowledge of the spatio-temporal forms of empirical objects does not depend upon the assumption that we have synthetic a priori cognition in mathematics (geometry and arithmetic), although he assumes that we do have such knowledge and that our a priori intuition of the pure forms of space and time explains it. He then infers from the premise that we have such a priori knowledge of the form of space and time and of the spatio-temporal form of objects that ‘Space [or time] represents no property at all of any things in themselves nor any relation of them to each other, i.e., no determination of them that attaches to objects themselves and that would remain even if one were to abstract from all subjective conditions of intuition’, and therefore that ‘Space [or time] is nothing other than merely the form of all appearances of outer sense, i.e., the subjective condition of sensibility’ (A26/B42; cf. A32–3/B49–50). Why does he make this inference? Because, he says, ‘neither absolute nor relative determinations can be intuited prior to the existence of the things to which they pertain, thus be intuited a priori’ (A26/B42). In one sense, this is trivially true: if to intuit something means to be given it, then of course I cannot intuit something before I am given it. But this would seem to apply only to the intuition of objects. When it comes to properties, it might seem that I could intuit them before being given any particular object that has them; to take an empirical case, given that I know that human beings can hear sounds only between the frequencies of 20 Hz and 20 MHz, I can know in advance of hearing it that any sound I do hear will fall within that range.
Kant seems to need a further argument for his claim that a priori intuition implies the subjectivity or mind-dependence of the properties intuited a priori. And he repeatedly offers such a further argument, the argument to which I earlier alluded when I noted Kant’s disanalogy between the subjective ideality of colours and the transcendental ideality of space and time. The argument is that what we know a priori we know to be necessarily true, and that we can know synthetic necessary truths only about our own representations, not about things that exist independently of our representations. Kant states this argument first in the General Remarks on the Transcendental Aesthetic. Here he starts from the example of a geometrical proposition supposedly known a priori and thus as necessarily true, the proposition that an enclosed figure can be constructed only with three straight lines, but the same argument could apply more generally to not specifically geometrical propositions, such as that two simultaneously existing objects can be intuited as distinct only by being intuited as located in different regions of space, or two events not located in different regions of space can be intuited as distinct only by being intuited in successive moments of time. Here is the argument:
If there did not lie in you a faculty for intuiting a priori; if this subjective condition regarding form were not at the same time the universal a priori condition under which alone the object of this (outer) intuition is itself possible; if the object (the triangle) were something in itself without relation to your subject; then how could you say that what necessarily lies in your subjective conditions for constructing a triangle must also necessarily pertain to the triangle in itself? For you could not add to your concept (of three lines) something new (the figure) that must thereby necessarily be encountered in the object, since this is given prior to your cognition and not through it. (A48/B65–6, emphasis added)
Kant’s argument is that, while spatio-temporality in general or the specific geometrical property he describes might be necessarily true of our representations, they would not be necessarily true of objects independent of our representations; but they are necessarily true throughout their domains, so they must not be true at all of objects of independent of our representations. Thus spatio-temporality and all that it involves must be only the subjective condition of representation.
The presentation of this argument in the Critique is no fluke. Kant repeats it in the Prolegomena (§13, note I, 4: 28) and in his unpublished notes (Reflection 5637, 18: 27). The claim is always that if the synthetic propositions about space that we know a priori and therefore know to be necessarily true were to hold of mind-independent things as well as of our own representations of them, we could only know them to be contingently true of the former, and therefore not know them to be necessarily true throughout their domain after all. But we do, so they must not be true of things in themselves at all.
This is Kant’s preferred argument for transcendental idealism, and it is his preferred argument precisely because it blocks the missing-alternative objection to which Allais’s reconstruction is, as we saw, open: it denies that spatio-temporal properties can be properties of things as they really are as well as of our representations of them because even though they would be necessary properties of the latter, we could at most know them to be contingent properties of the former; but we do know that spatio-temporal properties are necessary throughout their domain. Of course, even though this is Kant’s actual argument, it is not a very good argument, because there is an obvious objection to it. This is just that the possibility that spatio-temporal properties are properties of things in themselves would prove that we do not know them to be necessarily true throughout their domains after all. When faced with a modus tollens of the form ‘p implies q, not-q, therefore not-p’ – ‘the transcendental realism of spatio-temporality implies the contingency of our knowledge of it, but our knowledge is not contingent, therefore transcendental realism is false’ – we always have two choices: we can accept not-p or simply deny not-q, that is, just accept that our knowledge of spatio-temporality is not necessarily true throughout its possible domain. Of course, Kant does not countenance this alternative conclusion.
This completes my discussion of the argument for transcendental idealism itself. I close with the suggestion of a more general problem with Allais’s interpretation, namely, that I do not see how Allais can hold that her account is moderate because it is relational and yet insist that spatio-temporal objects are contained in and only in our intuitions of them. I have already conceded that the latter may be true of mathematical constructions in pure intuition, but on Kant’s view these are not actually determinate objects, and mathematics does not give us knowledge of objects at all except insofar as it is applied to empirical intuitions and their objects. As for the latter, Kant himself insists that empirical intuitions and their objects are distinct, although the empirical intuitions, unlike concepts, are our first and most direct representations of their objects. Allais’s moderate metaphysical interpretation of transcendental idealism begins by maintaining that mind-dependent representations give us mind-independent objects, although our mind-dependently spatio-temporal representations do not give us those objects as they are in themselves, which is non-spatio-temporal. Her containment-model of (empirical) intuitions conflicts with this relational model of perception, and for that reason as well as for its own lack of support ought not to be accepted. I think that we should accept Allais’s relational model of transcendental idealism, but her reconstruction of Kant’s argument for transcendental idealism is incomplete, and when we complete it, we should not accept Kant’s argument itself!