Introduction
Key to understanding Kant's transcendental idealism is his distinction between appearances and things in themselves. The debate about the proper interpretation of this distinction has for many years tended to be between two camps. One has held that Kant's distinction is ontological, a distinction according to which appearances are in some important sense in the mind, whereas things in themselves exist outside of it. The other has held that the distinction instead reflects two ways in which human beings can consider the world, ways that reflect no ontological distinction in the world considered. Considering the world as it appears might be to consider it as subject to conditions the representation of which permits objects to appear to us at all; to consider it as it is in itself might be to consider it in abstraction from the very features the representation of which allows it to appear to us in the first place.
Once a student of Kant could gloss the above positions as ‘two-world’ and ‘two-aspect’ readings, respectively, and hold that these exhausted our interpretative choices. No longer. ‘Two-world’ readings have lost their monopoly on construing appearances mentalistically.Footnote 1 Moreover, there are now ‘two-aspect’ readings that insist that the distinction between considering things as they appear and as they are in themselves does indeed track an ontological distinction.Footnote 2
All the same, understanding the old debate can help clarify what a successful interpretation of transcendental idealism must look like. The advantages of, and difficulties for, the views of these two camps have by now been well-documented.Footnote 3 But there is a simple way to account summarily for a lot of their strengths and weaknesses. Kant claims to be a transcendental idealist while also maintaining that he is an empirical realist about space and time (e.g. A27–8/B43–4, A35/B52).Footnote 4 A key problem for two-world readings is explaining the realism of empirical realism. On these readings, space, time, and appearances are located in the mind; but then Kant's claim to be any kind of realist about them is hard to take seriously. On the other hand, traditional two-aspect readings have difficulty explaining the idealism of transcendental idealism. What, after all, is the idealistic implication of being able to abstract from various apparent features (e.g. spatiality, temporality) of the objects that we experience? Certainly it does not follow that space and time are in the mind or that the apparent spatio-temporality of objects is merely apparent (two reasonable, albeit only partial, glosses of Kant's position that space and time are ‘mere forms’ of outer and inner sense, respectively). It becomes hard to see how appearances, space, or time should be in any meaningful way mind-dependent (see also Guyer Reference Guyer1987: 337–8; Van Cleve Reference Van Cleve1999: 4). An account of transcendental idealism thus faces a double imperative: to do justice to both Kant's realism and his idealism.
Really, this is old news. Lucy Allais, by contrast, is among those who have tried to meet the foregoing challenge in a new way. Allais's response places great emphasis on an analogy that Kant draws between appearances and secondary qualities. Ultimately, on her analysis, there is for Kant only ‘one world’, not two. It has this in common with two-aspect analyses of transcendental idealism, and this is key for how Allais accounts for Kant's empirical realism. On the other hand, the distinction between appearances and things in themselves is an ontological distinction according to which appearances are in an important sense mind-dependent. It has this in common with two-world analyses, and this is crucial for how Allais accounts for Kant's idealism.
Allais provides compelling criticisms of the familiar positions in this debate. I welcome her call to move beyond them. But in what follows I argue that, despite certain advantages that it can claim, her own reading is not the way to do it. In section 1 of what follows, I begin by briefly articulating Allais's proposal. I then consider, in section 2, certain points of detail about her reading that fall somewhat outside the scope of my initial presentation but are important for offering a balanced presentation of her view. In section 3, I explain why Allais's interpretation cannot be the right way to understand transcendental idealism.
1. Transcendental Idealism and the Secondary Quality Analogy
The passage in which Kant compares the ideality of appearances to the status that secondary qualities had long been thought to have occurs in the Prolegomena. Here we see Kant in the midst of recoiling at the suggestion that his idealism is akin to Berkeley's:
That one could, without detracting from the actual existence of outer things, say of a great many of their predicates: they belong not to these things in themselves, but only to their appearances and have no existence of their own outside our representation, is something that was generally accepted and acknowledged long before Locke's time, though more commonly thereafter. To these predicates belong warmth, color, taste, etc. That I, however, even beyond these, include (for weighty reasons) also among mere appearances the remaining qualities of bodies, which are called primarias: extension, place, and more generally space along with everything that depends on it (impenetrability or materiality, shape, etc.), is something against which not the least ground for uncertainty can be raised; and as little as someone can be called an idealist because he wants to admit colors as properties that attach not to the object in itself, but only to the sense of vision as modifications, just as little can my system be called idealist simply because I find that even more of, nay, all of the properties that make up the intuition of a body belong merely to its appearance; for the existence of the thing that appears is not thereby nullified, as with real idealism, but it is only shown that through the senses we cannot cognize it at all as it is in itself. (Prol 4: 289)
Allais thinks that Kant took the tradition to hold that secondary qualities are relational properties of mind-independent objects. Importantly, she does not, on behalf of this tradition, construe them as (1) illusory, (2) located in the mind, or (3) powers of mind-independent objects to produce certain representations in us.Footnote 5 For Kant's predecessors, as he saw it, secondary qualities are indeed properties that objects have independently of any particular mind. On this view, ‘objects have [e.g.] colour whether we are perceiving them or not, and would not change colour if we all died’ (Allais Reference Allais2004: 674). On the other hand, objects have secondary qualities, with their unique phenomenologies, only in relation to human minds in general: specifically, only in virtue of the fact that we can experience those objects (Allais Reference Allais2007: 464, 473–6; 2004: 671–6). This makes them, in an important sense, mind-dependent or ideal. If, then, an object has a secondary quality, it is a result of the object's particular mind-independent nature and the possible human experience of that object. The latter is, perhaps, a strange candidate for a relatum. But I do not think that scruples we might have about it will constitute a serious objection to Allais using this analysis to make sense of transcendental idealism.
Allais's proposal is this: Kant's claim from the Prolegomena is that all empirically real properties are relational properties in the above sense.Footnote 6 They are only the ways mind-independent things appear to subjects like us (Allais Reference Allais2007: 477; see also 2004: 672, 673); or, as Allais also puts it, they are ‘appearance properties’ of mind-independent objects (2004: 673, 674, 680 n. 29). Empirically real properties exist independently of any given mind or minds but not independently of some human mind possibly experiencing the objects in which they inhere. Appearances – things as they appear – are mind-independent objects together with the mind-dependent appearance properties that inhere in them.
Some of the examples that Allais uses to work up to the conception of secondary qualities that serves as her model for transcendental idealism have the potential to cause confusion about what appearance properties are supposed to be. So here I want to pause to add a note of clarification.
In order to bring into relief the conception of secondary qualities that she is after, Allais asks the reader to adopt a ‘relational view’ of perception. This view posits no representational intermediaries of any kind between the mind and the mind-independent object perceived: there is a perception of an object just when there is a sui generis relation between the mind and that mind-independent object. The point of this proposalFootnote 7 seems to be to undercut any temptation to identify secondary qualities with some mental intermediary or property thereof. These are not the conceptions of secondary qualities to which Allais thinks Kant appealed.
In the course of discussing the relational view of perception, Allais writes that
there is no reason to accept the claim that, when the stick appears bent in the water, there is anything that is bent; there is simply something that appears bent. While the stick does not have the property of being bent, it does have the property of appearing bent – the apparent bent-ness of the stick is a property of the stick, and not of a mental state or mental activity. (Allais Reference Allais2007: 472)
Shortly afterwards, Allais explains how the relationalist understands the stick having the property of appearing bent: ‘the appearing is something public’ (see also Allais Reference Allais2004: 671). The suggestion is that the stick appears bent even when no one perceives it. There need only be ‘the possibility of conscious minds’ for the stick to have the property of appearing to be bent (Allais Reference Allais2007: 472). This example is intended to get us comfortable with features of the account of secondary qualities to which Kant is supposed to have appealed: specifically, that they are public despite being mind-dependent.
The example, however, can easily mislead. It is tempting to infer that Kant's view is supposed to be that we can experience only the appearance of e.g. squareness and not squareness itself – just as we encounter only the appearance of bent-ness, not bent-ness, in the case of the stick in the glass of water.Footnote 8 And while it may be true in some sense that for Kant we encounter only the appearances of mind-independent things, it is not true that we encounter only the appearances of spatio-temporal properties. Indeed, the latter claim is exactly the sort of view Kant attacks in the Refutation of Idealism.Footnote 9
Fortunately, the foregoing is not Allais's position. It is not that we encounter the appearances of spatio-temporal properties in experience. Rather, we encounter those spatio-temporal properties. Allais's innovative claim is that these properties are appearance properties: ways that mind-independent objects appear to minds like ours (see e.g. Allais Reference Allais2007: 474). Squareness, for instance, is, and is nothing more than, a particular way that objects appear to our species (Allais Reference Allais2007: 477; 2004: 673).
The virtues of Allais's proposal are many. First, since her reading posits only one world of objects, it does justice to the numerous passages in Kant's texts that inspire two-aspect readings, passages in which Kant says that things as they appear are ‘the very same things as things in themselves’ (Bxxvii; see also Bxviii–xix n., Bxx, A38/B55, B69; Prol 4: 289). Yet, secondly, for Allais this distinction is indeed an ontological distinction, which acknowledges, what seems true, that Kant was after more with his distinction than two ways of considering the world. Third, Allais's interpretation seems to do justice to Kant's empirical realism: mind-independent things do indeed possess spatio-temporal features; these features are not located in any mind. And consequently Allais can offer a clear explanation of why spatio-temporality is not by Kant construed as illusory, a desideratum of any reading of transcendental idealism (B69–70, Prol 4: 287, 290–4, 374–5). For Allais, spatio-temporality is no illusion precisely because things really have spatio-temporal properties. They are not intrinsic properties of mind-independent objects. But they are still their properties (Allais Reference Allais2004: 674).
And since they are relational properties, where one of the relata is our possible experience of the objects that possess them, Allais's account explains – and this is a fourth virtue – the special mind-dependence, i.e. the transcendental ideality, of empirically real properties. Thus, things as they appear are in an important, but restricted, sense ideal. And as Allais notes, she thereby provides an elegant analysis of Kant's disapproval of ‘genuine idealism’ (Prol 4: 374–5). Kant writes:
Idealism consists in the claim that there are none other than thinking beings; the other things that we believe we perceive in intuition are only representations in thinking beings, to which in fact no object existing outside these beings corresponds. I say in opposition: There are things given to us as objects of our senses existing outside us, yet we know nothing of them as they may be in themselves, but are acquainted only with their appearances…. (Prol 4: 288–9)
Allais suggests that the idealisms from which Kant distances himself locate the objects of experience and their properties literally in the mind. But it remains open for Kant to hold that many properties of objects are mind-dependent in the way that Allais recommends: while they exist outside of human minds, their existence nevertheless depends on these minds, in that these minds must be able to experience the bearers of these properties (Allais Reference Allais2007: 463, 476–7; 2004: 676–7, 681).
If mind-independent objects are genuinely spatial and temporal, how does Allais explain Kant's position that ‘things in themselves’ are not spatial or temporal? For Allais, this is a misleading way to state Kant's claim. Rather, things as they are in themselves are neither spatial nor temporal. Even better: the intrinsic, mind-independent natures of things do not include spatio-temporal properties. This is consistent with these objects nevertheless being spatial and temporal.
2. Points of Detail
Despite the foregoing advantages of Allais's account, however, and despite my agreement with much that Allais writes, I think that we should resist her analysis. I want to present my reservations in two waves. First, I will present two concerns about elements of her reading on which I have so far not focused. In the next part of this paper I will directly address the interpretation that I discussed in section 1.
A single set of properties?
I have so far presented Allais as holding that all empirically real properties of objects are properties that objects have only in virtue of the fact that human beings can experience those objects. Allais takes seriously Kant's analogy to secondary qualities, and she takes Kant to understand secondary qualities in a particular way. Focusing on colour, she articulates the relevant account of colour as follows:
[T]he view of colour that we need is that it is a property of objects – and not of mental entities, mental states, or ways of mentally being – but that it is a mind-dependent property of objects. It is mind-dependent in the sense that the existence (or possibility) of minds is necessary for the existence (or possibility) of the property, but is not sufficient for it: it is a partially subjective and partially objective relational property of the object. (Allais Reference Allais2007: 464)
I have suggested that, for Allais's Kant, mind-dependent relational properties are one set of properties that things have and that their intrinsic properties are another – although the latter are partly responsible for the former.
Here, however, I want to acknowledge that the reading that I have presented seems not to be exactly Allais's considered view. For first, Allais sometimes says, of the view of secondary qualities that she appeals to in her analogy, that it is an object's mind-independent properties, rather than the object, that appear coloured. Indeed, she goes so far as to describe colour not as a property of an object but as ‘the mode of presentation of a [mind-independent] property’ (Allais Reference Allais2007: 476 n. 77; see also 479–80; and 2004: 679–80). Second, Allais seems to hold that, in fact, transcendental idealism does not ultimately rest on a distinction between two sorts of properties, but on a distinction between ‘two ways of knowing the same [intrinsic] propert[ies]’.Footnote 10 These features of her position, I suggest, do not cohere with the rest of her analysis. Once they are excised, what remains is the position that I have sketched in section 1.
In a way, the difficulty that Allais here introduces is the same difficulty that Kant is often said to face. There is reason to think that Kant holds that appearances and things in themselves are distinct things. For instance, the former are said to be spatio-temporal, the latter not. On the other hand, there is also reason to think that he holds that appearances are not distinct objects from things in themselves. As I mentioned above, Kant often says that they are identical. Allais faces the difficulty of holding that appearance properties are mind-dependent and relational and yet, at the same time, holding that they are identical to properties that are mind-independent and non-relational. Part of what seemed exciting about the view that I sketched in section 1 was that it helped us escape from this kind of difficulty. Allais, however, reintroduces it in a different guise.
If Allais commits to this perspectival component of her view, then one of the difficulties that she runs into is explaining how she avoids saddling Kant with the position that what appears to us is systematically illusory. In response to this concern about the account of secondary qualities that she considers, Allais writes:
Such ways in which objects appear [e.g. coloured], or modes of presentation of objects, cannot be thought of as misrepresenting a property of the object in itself, because in these cases the mode of presentation does not give a distorted presentation of a property the object has in itself, but rather presents a property which the object does not have apart from its perceptually appearing to us. (Allais Reference Allais2007: 474)
This passage is understandable if we assume that colours differ from an object's intrinsic properties. But – now returning to her reading of Kant and not simply the secondary quality analogy – Allais holds (as she must) that an object's intrinsic properties are neither spatial nor temporal. If those properties appear to us to be spatial and temporal, then that is, trivially, misrepresentation. Accordingly, she foregoes her account's ability to explain the empirical reality of spatio-temporal properties, an ability I have billed as one of its virtues.
I do not mean to suggest that it is unmotivated for Allais to claim that the appearance/thing in itself distinction ultimately rests on two perspectives that we can take up on objects, or on their properties. There are, as I will argue, problems with the view that I sketched in section 1, and a traditional two-aspect interpretation might have the resources to avoid some of these objections. My claim is rather that Allais cannot press the secondary quality analogy into service in the way that she does without also holding that appearance properties are really distinct from intrinsic properties. Since the analogy to secondary qualities is clearly more fundamental to her position, I suggest that she must abandon the claim that transcendental idealism turns on two ways of knowing a single set of properties.
The relational view of perception
Above I said that Allais appeals to a relational view of perception to bring into relief the view of secondary qualities – that they are special kinds of relational properties – that she believes serves as a model for Kant's idealism. Accordingly, we might take this appeal to be a heuristic device and nothing more: in particular, we need not construe Allais to be proposing that Kant embraced a relational view of perception. About Kant's theory of perception, Allais could be agnostic.
There are certainly passages that suggest that Allais is not trying to attribute to Kant a view of perception, let alone a relational one (Allais Reference Allais2004: 670; 2007: 468). Yet there are also passages that suggest that she is doing precisely this. For instance, at one point Allais suggests that if adverbial, intentional object, or sense-datum theories ‘were used to present Kant's idealism about appearances, it would … reduce to a phenomenalist idealism, which would mentalize appearances’ in a way she has already rejected, even if adverbial and intentional object accounts would not construe appearances ‘as mental objects’ (Allais Reference Allais2007: 469). The implication seems to be that only a relational view of perception is compatible with the account of appearances that she wants to attribute to Kant.
Perhaps Allais would insist that she here meant only to claim that these other views of perception could mislead about the account of secondary qualities on which she wants to model spatio-temporal properties generally. But later she considers objections to her account of transcendental idealism that object to her attributing a relational view to Kant. Her responses seem to acknowledge that she is indeed making this attribution.Footnote 11 (In more recent work, Allais unequivocally imputes a relational view of intuition to Kant.Footnote 12)
I am inclined to think that Allais need not do this. There is no obvious tension in attributing to Kant the view that the properties that appear to us are special kinds of relational properties, while also attributing to Kant, say, an intentional object theory of perception. Consequently, I do not want to press too hard objections that follow from trying to fit a relational view of perception on to his thinking. On the other hand, this would be a very novel interpretation of how Kant conceives of the mind–world relation, and I want briefly to consider whether and why it really is unacceptable.
If one is to attribute a relational view of perception to Kant, one must explain where intuitions fit on this picture. At first blush, they seem not to. It is natural to construe intuitions as the direct objects of perception (a sense-datum reading) or as the representational vehicles – not themselves perceived – of objects appearing to be in space and time (an intentional object reading). Both construals are incompatible with a relational view, which denies any sort of perceptual intermediary.Footnote 13
There may be another way to understand intuitions, however. The advocate of a relationalist reading of Kant might argue that for Kant an intuition – or at least an empirical intuition – is a complex consisting of the mind, the mind-independent object intuited, and the unique intuitional relation between them. This makes, as Allais notes (2010: 60, 65; see also 2007: 468), the object intuited an essential ‘constituent’ of the intuition. And while I do not endorse this proposal, for the sorts of reasons that I will emphasize further below, it does make ready sense of Kant's occasional claims that an intuition ‘is dependent on the existence of the object’ (B72).Footnote 14 After all, if an intuition is in part constituted by a mind-independent object, then of course the existence of the intuition would depend on that object.
Relational views, however, have implications. Prima facie, they cannot explain hallucination and dreaming. For these are cases in which no mind-independent object plays the role of constituent: the pink elephant that I am hallucinating, or the dragon about which I am dreaming, does not exist. And it has seemed to many philosophers that hallucinations and dreams must form a ‘common kind’ with veridical perceptions.Footnote 15 This is impossible if hallucinations and dreams cannot also be explained relationally.
The only way forward appears to be to construe Kant as a disjunctivist. Disjunctive accounts of perception deny precisely that hallucinations and dreams form a common kind with veridical perceptions. A disjunctivist might thus have a relational view of perception but e.g. a sense-datum account of hallucination. And indeed, in the context of explaining her account of appearances, Allais claims that a relational view must be disjunctivist (Allais Reference Allais2007: 469) – although she does not explicitly attribute disjunctivism to Kant.
This would be one of the novel implications of reading Kant as holding a relational view of perception. And, in fact, I think that there may be some textual basis for supposing Kant to be a disjunctivist. Disjunctivists do not deny that veridical experiences and hallucinations can share the epistemological property of being (in principle) subjectively indistinguishable. But they deny that there is deeper non-epistemological commonality that explains the indistinguishability, such as e.g. qualitatively identical sense-data. One way that a disjunctivist could mark this point is by refusing to use some favoured term – such as ‘experience’ – for cases of hallucination and the like to begin with. Interestingly, Kant consistently holds that hallucinations and dreams (‘imagination’) are not experiences (Bxl n., B275, B278–9; see also A202/B247).
Yet overall it really does appear to be Kant's position that veridical experiences, hallucinations, and illusions form a common (i.e. non-disjunctive) kind. Kant seems to hold that ‘dreams as well as… delusions’, along with veridical experiences, are in part constituted by ‘intuitive representation[s]’ (anschauliche Vorstellung).Footnote 16 In the Anthropology, he writes that cases of imagination employ the faculty of intuition ‘without the presence of an object’.Footnote 17 So, dreams, hallucinations, and veridical experiences seem to have intuitions, or at least ‘intuitive representations’, in common. There is not much reason to think that Kant was a disjunctivist, and passages suggesting that an intuition depends on the presence of an object can be read as Kant simply emphasizing the passivity of our faculty of intuition, in contrast to the understanding.Footnote 18 But if Kant was no disjunctivist, then hallucination poses a genuine roadblock to attributing a relational view of perception to him.Footnote 19
3. Reasons for Doubt
As I said, there is at least a little ambiguity about whether Allais means to attribute a relational view of perception to Kant, and (in my opinion) even more about whether she needs to. But the reading of transcendental idealism that I presented in section 1 clearly comes with other problematic commitments. I turn to those – three of them – now.
Appearances as mere representations
For Allais's Kant, the properties of objects that we experience are appearance properties, which are relational properties – properties possessed by objects only by dint of their bearers being possibly experienced by us. Allais considers it a virtue of her reading (a point that I second) that she can make these features mind-dependent without literally placing them in the mind.
The trouble is that Kant often suggests that appearances are in the mind or are ‘mere representations’. For instance:
[W]hat we call outer objects are nothing other than mere representations of our sensibility … (A30/B45)
For as appearances they [i.e. ‘the objects with which our cognition has to do’] constitute an object that is merely in us, since a mere modification of our sensibility is not to be encountered outside us at all. Now even this representation – that all these appearances and thus all objects with which we can occupy ourselves are all in me, i.e., determinations of my identical self – expresses a thoroughgoing unity of them in one and the same apperception as necessary. (A128, A129)
Space itself, however, together with time, and, with both, all appearances, are not things, but rather nothing but representations, and they cannot exist at all outside our mind … (A492/B520)
Such passages are a major motivation for classical two-world readings, and they are hardly isolated instances. Allais's reading makes poor sense of them.
Or so I say. Allais is mindful of this concern. And while she does not address passages suggesting that appearances are ‘in’ our minds, she does suggest that her analysis can account for Kant claiming that appearances are representations:
Like the idea of non-veridical perception [e.g. the bent stick illusion], the secondary quality analogy enables us to make sense of Kant's calling appearances ‘representations’. So long as there is a possible view of perception which allows that perception can be directly of external objects, yet not necessarily reveal the way they are in themselves, we can make sense of saying that we perceive objects as they appear, or appearances of objects, and since these may be different from the way objects are as they are in themselves, we can see the point in saying that they represent objects. The term stresses the fact that perception involves objects appearing for subjects, and that the way things appear presents, or represents, things as being a certain way for subjects. (Allais Reference Allais2004: 674; see also 671; and see 2007: 473)
Allais's point seems to be that appearance properties – which are mind-dependent but inhere in mind-independent objects – are ‘representations’ in a perfectly intelligible sense of this expression: they present to us the objects that bear them.
But this is not a plausible gloss of Kant's position that appearances are mere representations. There may be some philosophical point to calling appearance properties – if we choose to countenance such things – representations, or presentations, of the objects of which they are properties. But can we reasonably attribute this innovation to Kant? For Kant, representations (Vorstellungen) include such things as intuitions, concepts (conceivings), judgements (judgings), cognitions, perceptions, and sensations (e.g. A320/B376–7). All of these are supposed to be located within the mind.
I want to be careful here. I do not mean to suggest that we must embrace a phenomenalistic or two-worlds interpretation of transcendental idealism according to which appearances are mental entities. That is one way to account for Kant's claims that appearances are ‘in’ us. But it is not the only way. If a paranoid person fears (let us assume wrongly) that he is being followed by assassins, we might say that the assassins are only ‘in his head’. But by this we do not mean that they literally exist within his skull. More likely we mean that he believes or otherwise represents assassins chasing him whereas in fact none are. So, too, we might be able to account for Kant's idealism by holding that we represent the world spatially and temporally when really, in itself, it is neither spatial nor temporal. This would be to construe spatio-temporal appearances as intentional objects of a certain sort; the idealism would consist in the claim that we are responsible for much of the way those objects appear.Footnote 20 To be sure, if we take seriously Kant's claims that appearances are identical to things as they are in themselves, then such an analysis faces the danger of collapsing into the position that our experiences are thoroughgoing illusions. I do not mean to minimize the seriousness of this danger. My point for present purposes is simply that such an analysis of Kant can do justice to Kant's phenomenalistic-sounding rhetoric without placing appearances literally in the mind. For present purposes, it is here, and only here, that I am drawing a line in the sand: the seeming phenomenalism must be explained, whether by a classical two-world reading, a reading according to which appearances are intentional objects, or some other. Allais has an account of the mind-dependence of appearances. But it fails to do justice to that component of Kant's thought that has led so many commentators to think that Kant ‘is closer to Berkeley than he acknowledges’ (Strawson Reference Strawson1995: 22). To borrow from Allais: her account of transcendental idealism is in fact ‘not idealist enough’ (2004: 656).Footnote 21
Idealism and imposition
A related point is this. There is a long tradition of reading Kant as, in some sense or other, holding that we impose spatio-temporality on to the content of our experiences.Footnote 22 No doubt this is in part because Kant says that we ‘put into’ (in…. legen) the objects of our experience features such as spatio-temporality (Bxviii).
Traditional interpretations of transcendental idealism can straightforwardly explain this claim. It is worth emphasizing how. Advocates of readings according to which our experience of spatio-temporality is a systematic illusion can say that we introduce the (mis)representation of space and time into our experiences of things in themselves. Advocates of two-world readings can say that we very literally introduce space and time into empirical reality since, it turns out, empirical reality is mental to begin with, and there is no special difficulty about how we can affect the goings-on of our own minds. It is, on the other hand, difficult to see how Allais can explain the imposition of space and time on to appearances.Footnote 23 Precisely because Allais understands spatio-temporal features to be properties that mind-independent objects really have, there seems no imposition. Granted, mind-independent objects have these features only because we can possibly experience these objects. But this is no imposition. Queen Elizabeth II possesses the property of being a wife only in virtue of the existence of Prince Philip. But presumably the prince did not impose wifehood on to her.
Yet maybe this criticism rests on the sort of outworn understanding of transcendental idealism that Allais is asking us to abandon. Perhaps we ought not to suppose that by ‘putting into’ appearances such features as spatiality and temporality Kant has in mind anything like an imposition.
It is hard, however, to accept this recommendation. One traditional reason to hold an imposition model is that Kant thinks that we represent spatio-temporality through a priori intuitions. Part of what makes these special is that they are supposed to be our contribution to the experience of objects: imposed, not passively received from without, as is the case with empirical intuitions. ‘It is clear from what has been said above … that the first condition, namely that under which alone objects can be intuited, in fact does lie in the mind a priori as the ground of the form of objects’ (A93/B125; my emphasis, except for a priori). Moreover, Kant does not say that we ‘put into’ the objects of experience various formal features only in the B-Preface. He also makes this claim toward the end of the A-Deduction:
Thus we ourselves bring into (bringen wir selbst hinein) the appearances that order and regularity in them that we call nature, and moreover we would not be able to find it there if we, or the nature of our mind, had not originally put it there (hineingelegt). (A125; see also A114, A127, B159–60; Prol, 4: 319)
Indeed, it seems an important aim of both editions of the Transcendental Deduction to explain how the forms of the understanding (the forms of judgement), together with those of the sensibility (forms of intuition), structure appearances in the ways ultimately described in detail in the Analytic of Principles. This structuring occurs via acts of ‘synthesis’. And synthesis involves a kind of imposition.
Admittedly, such a reading of the Deduction, even this vaguely stated, is not entirely uncontroversial, for it implies that Kant is in the business of making, broadly speaking, psychological pronouncements. For many, Kant is in no position to make such pronouncements, either because so construed Kant's claims would be simplistic armchair theorizing (Young Reference Young1994: 332) or because they would be based on empirical observation and thus unable to help justify a priori principles (for discussion see Kitcher Reference Kitcher1990: 6, 23–4). Consequently, many sympathetic commentators have urged readers to interpret passages, or at least some passages, that appear to dabble in ‘transcendental psychology’ to be, fundamentally, epistemological claims.Footnote 24 And Allais may have some sympathy with this line of thought:
In his summary presentation of the argument from a priori cognition in the B preface [of the Critique of Pure Reason], Kant suggests that the argument applies in a parallel way with respect to a priori intuitions and a priori concepts; correspondingly, transcendental idealism is frequently thought of as involving the idea that our a priori forms of intuition and our a priori categories play a similar role in structuring the objects of experience. [Allais's analysis of Kant's argument in the Transcendental Aesthetic] questions the tightness of this parallel…
Allais goes on to claim that considerations about the nature of intuitions and concepts – considerations that I cannot enter into here – suggest
the possibility of a reading of transcendental idealism according to which the idealism (however this is understood) is fully established in the Transcendental Aesthetic, while Kant's concern in the Deduction is simply to limit the categories to whatever kind of mind-dependence the a priori forms of intuition involve. (Allais Reference Allais2010: 71)
Without question, this is a goal of the Deduction.Footnote 25 But to bill it as Kant's exclusive goal requires ignoring many important passages in which Kant really does seem to claim that the understanding or imagination contributes to structuring appearances. Kant tells us that syntheses of the imagination – ultimately we learn that these syntheses are in some way categorial (e.g. A125) – are required even for ‘the purest and most fundamental representations of space and time’.Footnote 26 ‘The imagination is … also a faculty of a synthesis a priori … [I]t is only by means of this transcendental function of the imagination that even the affinity of appearances, and with it the association and through the latter finally reproduction in accordance with laws, and consequently experience itself, become possible’ (A123; see also A118). The synthesis of the imagination, ‘which is an effect of the understanding on sensibility’ (B152), ‘determine[s] the form of sense a priori’ (B151–2) and thereby, Kant seems to argue, prescribes categorial laws ‘to nature’ (B159). This structuring is importantly like the structuring provided by our forms of intuition: ‘It is by no means stranger that the laws of appearances in nature must agree with the understanding and its a priori form, i.e., its faculty of combining the manifold in general, than that the appearances themselves must agree with the form of sensible intuition a priori’ (B164). This happy result addresses Kant's worry at the outset of the Deduction that ‘appearances should contain anything’ that answers to ‘the concept of cause’ (A90/B122) or to the other categories: the categories, Kant concludes, are indeed objectively valid.
Thus, limiting the categories to ‘whatever kind of mind-dependence the a priori forms of intuition involve’ cannot be Kant's sole agenda in the Deduction. And I hope that the above passages suggest, further, how difficult it is to deny that spatio-temporality is for Kant in some way or other imposed on to appearances by the mind in a way that Allais's interpretation seems unable to explain.Footnote 27
Space and time
In the previous objection, I appealed to the tight connection that Kant sees between spatio-temporal representation and the understanding. If the understanding is imposing categorial structure on to appearances, it is very awkward to suppose that the sensibility is not likewise imposing spatio-temporal structure. But there is an even more fundamental connection to attend to: the relationship between spatio-temporal properties and space and time themselves. This yields a final objection to Allais's account of transcendental idealism.
Allais provides a reading of the transcendental ideality of properties that appear to us in experience. These are, of course, spatio-temporal properties. Again, her claim is that these are mind-dependent properties that genuinely inhere in mind-independent objects. Allais says very little, however, about the ideality of space and time. How could her analysis account for their ideality?
Can it be that space and time – those ‘extraordinary particulars’ (Guyer Reference Guyer1987: 346) in which all spaces and times are located (A24–5/B39, A31–2/B47) – are mind-dependent properties of mind-independent objects? There is a philosophical problem with this suggestion: there seems no non-arbitrary mind-independent object to which we could assign space and time. There are also textual difficulties. It is just not feasible to square this account of the ideality of space and time with Kant's claims that ‘[s]pace 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 …’ (A26/B42; see also B70) and that ‘[t]ime is nothing other than the form of inner sense’ (A33/B49).
Perhaps, then, the ideality of space and time must be understood differently from the ideality of spatio-temporal properties?
One might try to argue that space and time are ontologically secondary to spatio-temporal properties. Space and time themselves might be entities that arise in virtue of the existence of spatio-temporal properties but without inhering in the same objects in which those properties inhere. But this recommendation really comports no better with the idealist passages that I quote above. Moreover, it is anyway implausible to suggest that Kant held that spatio-temporal properties are ontologically prior to space and time. Clearly Kant thinks that spatio-temporal properties are representationally secondary: our representations of them are possible only because we first have (a priori) representations of space and time (A23–4/B38–9, A30–1/B46). More importantly, if it makes sense to talk about ontological priority here, it seems that Kant would have to hold that space and time are what are prior. For instance: he holds that we can know a priori that empirical objects have many of the extended properties that they do because they are in space, which constrains them to obey the laws of Euclidean geometry.
Instead maybe Allais would simply hold that, while we are presented with spatio-temporal properties that mind-independent objects genuinely have, space and time themselves are merely intentional objects: we represent objects as in space and time when really they are not.Footnote 28 But it cannot be right that for Kant spatio-temporal properties exist outside of space and time. Indeed, it is worth stressing that the ideality that Kant attempts to establish in the Transcendental Aesthetic is explicitly the ideality of space and time. Kant includes additional arguments in the B-edition (B66–72), but plainly he feels entitled to conclude that the objects of experience are ideal precisely because they are spatio-temporal and because he thinks he has already established the ideality of space and time.Footnote 29 This implies that the current ‘out’ that I am considering on Allais's behalf will not work: Kant is committed to keeping spatio-temporal properties within space and time. More importantly, it suggests that in fact the ideality of space and time must be understood in the same way as the ideality of spatio-temporal properties.Footnote 30 If, as a reading of Kant, it is unacceptable to make space and time properties of mind-independent objects, it must be equally unacceptable to make spatio-temporal properties genuine features of mind-independent objects.
As I have said, I endorse Allais's ambition of charting ‘middle course’ between two-world and traditional two-aspect readings of transcendental idealism (Allais Reference Allais2007: 460; and see 479). She has made quite clear the need to do so. But I recommend against her ingenious, positive proposal.Footnote 31
Email: andrew.roche@centre.edu
