I am extremely grateful for having such excellent, interesting, challenging and varied commentators, and for their having read and engaged with my work in such depth.

Manifest Reality: Kant’s Idealism and his Realism presents an interpretation of Kant’s transcendental idealism. I have a few aims. One is to find a way of understanding Kant’s position that does justice to his being an idealist – his holding that physical objects in space and time depend on our minds, in some sense and to some extent – at the same time as accommodating his explicit rejection of understanding this mind-dependence as anything like a Berkeleian idealism which sees physical objects as existing as constructions out of what exists merely in the mind. Further, I aim to do this in a way that accommodates Kant’s holding that the things that appear to us have a way they are in themselves, independently of us, that grounds the way they appear to us, and of which we cannot have knowledge. And I aim to present an interpretation that illuminates the connections between transcendental idealism and Kant’s account of cognition, with respect to both empirical and metaphysical cognition.

A central challenge is to find space for a kind of mind-independence that does not involve existence literally in the mind. My attempt to do this draws on the analogy with so-called secondary qualities like colour that Kant presents in the Prolegomena, where he says that his position makes all the qualities of things that we experience relevantly like colour in being mind-dependent appearance qualities, but that this does not detract from the actual existence of these outer things or amount to Berkeleian idealism (4: 289). There are many different philosophical accounts of colour, which means the analogy could be read in many ways. I aim to present an account that enables us to make sense of Kant’s idealism in a way that meets the above desiderata. To this end, I draw on a range of philosophical accounts of perception that have been called relational or direct realist. These positions see the object perceived as a constituent of a perceptual mental state, as opposed to those which hold that perception involves being in immediate mental contact with something merely in the mind. I argue that what we need to make sense of Kant’s position is a view of perception which allows that perception involves objects outside us as constituents of perceptual states, but which also allows that direct perceptual states may present objects as being other than they are independently of their being perceived. I argue that it is within this kind of position that we can situate the account of colour that we need to understand Kant’s secondary quality analogy. As I read him, he holds that all the qualities of objects given to us in perception are mind-dependent ways in which things are presented to us, which involve the things themselves (not merely mental ideas representing these things), but do not present these things as they are independently of their being presentable to us.

Both Paul Guyer and Andrew Stephenson have worries about the account of perception to which I appeal in order to present the secondary quality analogy. Guyer is concerned about whether the view I appeal to is a possible view, saying that a relational account would see space and time as depending on a relation to the subject, whereas a direct realist account would suggest that spatio-temporal properties are real properties of things in themselves that are manifested to us in perceptual experience. In philosophy of perception there is no unproblematic or uncontested terminology and there are a number of ways in which different accounts could understand perception as relational. My use of this term draws on those contemporary philosophers of perception who use it to mark the idea that perceptual mental states involve the objects perceived as constituents, which means these views have an overlapping centre with direct realism. I draw on what is common to such accounts in order to present a possible position according to which we are directly presented with properties of things which they have only in our possible perception of them.

Like Guyer, Stephenson has worries with the account of perception I draw on. Stephenson is worried that the relational account of perception I use to present the secondary quality analogy is in fact incompatible with the relational account of colour the analogy requires. He argues that in contemporary philosophy of perception, the motivation for relational accounts of perception and the motivation for relational accounts of colour go in reverse directions. My initial response to this is to say that there may be many different motivations for both views; these are not my concern. Rather, my aim is to draw on a possible account of perception, to use it to illustrate a possible account of colour, and then to say that this account of colour gives a way of making sense of the kind of mind-dependence (and kind of mind-independence) that Kant attributes to all the properties of objects that we can experience. However, Stephenson argues that seeing the reverse motivations for the view shows why they are not actually compatible. Note that even if this were true, it would not necessarily undermine my view as an interpretation of Kant – it could be a problem for Kant’s position. But I am not convinced by Stephenson’s reasoning.

He argues that the notion of constitution that features in the relational view of perception must be understood as saying that what it means for a property to be perceived is for an object that has this property to be present in a perceptual state. On this account, the phenomenal qualities just are properties of the object: the redness in a visual perceptual state is the redness of the object, and it is in this sense that the objects are literally constituents of the perceptual experience. Stephenson argues that with relational accounts of colour, the order of explanation is exactly reversed: what it means for an object to be coloured is understood in terms of colour’s being present in a perceptual experience. We can’t have it with respect to the same properties, he argues, that that what it is for a property to be perceived is explained in terms of its belonging to an object that is present in a perceptual mental state and that what it is for an object to have that property is for it to be possibly perceived.

There are many ways of motivating and explaining the idea that the object perceived is a constituent of a perceptual mental state and they do not all appeal to phenomenal character. A common strategy is to explain it through denying that a perceptual mental state and a merely subjectively indistinguishable mental state that does not involve the presence of the object are metaphysically identical states – by rejecting a certain common factor view. In my view, this leaves it open how we should understand phenomenal character and what facts ground phenomenal character. Stephenson’s account seems to me to hold that direct realism has to be naïve realism: holding that objects are constituents of perceptual mental states would commit us to holding that objects must be perceptually presented as being just as they are independently of their being perceived in all respects. I think these views are separable. I agree with Stephenson that we cannot explain what it is to perceive some specific property in terms of that property’s being a perception-independent feature of an object that is presented in a perceptual mental state and at the same time explain what it is to have that property in terms of its being possibly presented in a mental state. But I think that explaining perceiving a property in terms of its belonging to an object present in perception is different to giving an account of the nature of the property. And I think that holding that an object is directly present to us in perception leaves open both the extent to which it is presented to us exactly as it is independently of its being perceived and the extent to which what explains the properties the object is experienced as having could include features of the object itself that are not presented to us. Stephenson does not think this is a possible view, but it seems to me to be exactly Kant’s view.

Central to my reading of transcendental idealism, and closely related to the direct realist account of perception I draw on, is my understanding of Kant’s key term ‘intuition’. Kantian intuitions are singular and immediate representations that give us objects, in contrast with concepts which are mediate and general representations that cannot give us objects. I understand the singularity and immediacy of intuition to mean that intuitions present us with particulars, where ‘presentation’ involves the actual presence of the object represented. And I read Kant as holding that intuitions do not depend on the application of concepts to play their role of presenting us with the objects of cognition. On my reading, Kant holds that cognition would not be possible without acquaintance with the objects of cognition, and this is what intuition gives us. This plays a central part in my account of Kant’s argument for the ideality of space and time. In my view, it is because he holds our representations of space and time are both intuitions (give us immediate presentations of what they represent) and a priori (are present to us without anything affecting our senses) that he holds that they could not present mind-independent features of reality.

Holding that intuitions do not depend on conceptualization to present us with particulars is centrally important for reading the Transcendental Deduction of the Categories, and, in particular, Kant’s crucial notion of synthesis in this argument. As I read Kant, the categorial synthesis he argues for in the Deduction is not something that is done to sensations in order to produce intuitions (to immediately present us with particulars), but something done to intuitions in order that they can be cognized. I see the Deduction as containing an epistemological argument for the claim that applying the categories is a condition of referential empirical concept application. Kant then is able to convert this conditional claim about those objects we can cognize to a claim about all objects in space and time because he has already established that objects in space and time are limited to the conditions of our cognizing them.

Guyer has a number of concerns about the role of intuition in my account and how it relates to Kant’s argument for his idealism. He accurately presents my view that Kant’s argument for the ideality of space and time follows from combining the notion of intuition with the notion of apriority. In my view, this leads to ideality, because Kant holds that the only way things independent of us get to be immediately present to our consciousness is by affecting our senses, and he holds that our representation of space presents its ‘object’ immediately (it is an intuition) but without affecting our senses (it is a priori). It therefore does not present something mind-independent. Guyer points out that pure intuition does not present fully determinate objects at all and argues that for this reason it cannot involve the direct presence of any object. To an extent I agree, because space and time, for Kant, are not objects (so not fully determinate objects) but forms; the forms in which empirical objects can be presented to us. However, as I understand him, Kant does think that these very forms are immediately present to us, both in experience of spatio-temporal objects and in pure intuition. Guyer worries that my version of the argument would, implausibly, make it rest merely on the definition of intuition. However I do not hold that Kant takes the ideality of space to follow from the definition of intuition alone, but rather from the implications of this definition for what it means for an intuition to be a priori (what an a priori intuition can present us with), together with arguments for the claim that our representations of space and time are a priori intuitions.

Guyer holds that Kant argues for the transcendental ideality of space and time on the basis of necessity: that fundamental propositions about space and time, including those in mathematics, are necessarily true of their objects. He argues that this reading does, whereas mine does not, block the so-called neglected alternative: the possibility that space and time could be the forms of our intuition and also features of mind-independent reality. The sense in which I take this possibility to be blocked is as follows: because of the way he understands ‘intuition’, holding that our representations of space and time are a priori intuitions means for Kant that what is presented to us in these representations (the space and time present to us in our experience of objects and in pure intuition) is not a feature of mind-independent reality. There could be something correlated with space and time in mind-independent reality, but this something is not what is presented to us in our representations of space and time, and it could not be, because then it would not be both a priori and an intuition. Thus a significant version of the neglected alternative is excluded and, unlike Guyer’s version of the argument, my version exactly tracks the way Kant presents it, and the point at which he takes his conclusion to follow: he presents arguments for the claim that our representations of space and time are a priori and intuitions, and immediately concludes that they do not present us with mind-independent features of reality.

Sasha Newton’s comments primarily focus on the non-conceptualist part of my reading of intuition and the implications of this for a Copernican account of knowledge. Newton is happy to admit non-conceptualism in the sense of the bare possibility of having intuitions without concepts, but argues that there is a significant sense in which intuitions are not separable from conceptualizing when I have knowledge of an object. There are different things that could be understood by ‘separable’ here, including the idea that we could introspect and separate out an unconceptualized part of intuition, and the idea that we have some actual intuitions that are not conceptualized. Neither of these are the kind of independence with which my non-conceptualist reading of intuition is concerned. Rather, my view is that intuitions do not depend on conceptualization to play their role of giving us acquaintance with particulars (giving us objects), which is compatible with it being the case that all our actual intuitions are conceptualized, and that we would be incapable of separating out an unconceptualized part of our intuition. Newton argues that although the roles of intuition and concepts are distinct, they cannot be understood apart from one another and cannot play their distinctive roles in an act of knowing independently of each other. I agree with this in the sense that, if sensibility gave an object independently of conceptualizing, that object could not be known or cognized. However, in my view this simply amounts to saying that being given an object is not enough for cognition or knowledge, not that being given an object depends on conceptualizing.

Newton attributes to me a view of the separability of intuition and concepts according to which knowledge is a loose unity of intuition and concept, on the model of an artefact in which a plan (concept) could have transformed the same raw material into something different. However, while I do insist on the independence of intuitions from concepts in terms of their playing their role in giving us objects, I do not think of intuition as a raw material that could be transformed into different objects if different concepts were applied to it; I think of intuitions as presenting us with the very objects about which concepts enable us to make judgements.

I discuss intuitions and concepts as ingredients of cognition (Erkenntnis), understood as something distinct from knowledge (Wissen), where the conditions for cognition concern what it takes to have a certain kind of (not necessarily factive) objective representation, which is a separate question from the kind of justification required for knowledge. Newton points out that Kant’s project about the possibility of metaphysics surely concerns knowledge of metaphysical claims, and says that we cannot show how these truths are possible without showing how we can have knowledge of them. I read Kant’s ‘how possible’ question as concerned with something other than justification: when he asks how synthetic a priori propositions are possible, I see him as asking how it is possible that such propositions could possibly qualify as cognitions. The worry here concerns how it can be possible that they meet one of the requirements of cognition – that of being about objects with which we can have acquaintance – since they are a priori and the ordinary way in which we have acquaintance with objects is through their affecting our senses. This is why Kant’s first answer to his ‘how possible’ question is the idea of a priori intuition: a representation which gives us acquaintance with something (because it is intuition) but does this independently of anything affecting our senses (because it is a priori). In my view, his account of how we justify synthetic a priori propositions (through proofs in pure intuition, in the case of mathematics, and transcendental arguments, in the case of metaphysics) is subsequent to his having answered the initial ‘how possible’ question, not a part of the initial answer, as Newton seems to think.

Newton argues that my reading leaves Kant with an account of knowledge according to which it could be in a significant sense accidental that one’s responsibly formed beliefs are true. This seems to be because she thinks that only the shaping of the object of knowledge by the understanding would result in a non-accidental correlation. I do not take Kant’s account of what it takes for us to have acquaintance with the objects of cognition as an account of what guarantees how reasonably formed beliefs are not accidentally true. On the other hand, I do think it plays an important part in a picture concerning knowledge, and the fact that intuitions, on my account, actually present their objects, rather than merely representing or presenting ideas about these objects, is crucial. This part of the picture is not, however, about how we stand in the space of reasons, since, as I read Kant’s account, he does not take merely having acquaintance to put us in the space of reasons.

Finally, Newton objects that my account of how Kant shows that the categories necessarily apply to all spatio-temporal objects involves an illicit slide from the claim established in the Aesthetic that objects that can be given to us are limited to our a priori intuition ‘to the more general claim that they must be possibly cognized, or are limited to conditions of cognition’ (p. 280). What is crucial, as I understand Kant here, is the final stage of the Deduction, where he argues that we need to synthesize space and time in ways governed by the categories in order to cognize them. He says that it follows that everything given in space and time is subject to this synthesis: everything given in space is subject to categorial synthesis. This, as I read him, is what enables him to conclude that what is given in intuition is not just limited to the conditions of intuition but also to those of the conceptual component of cognition.

I argue at some length against what I call phenomenalist interpretations of transcendental idealism (largely in chapter 2). However, Nick Stang argues that the view I present is not as different from phenomenalist interpretations as I see it as being. There is a way in which I think this is right. As Stephenson notes, the model of perception I draw on to introduce the secondary quality analogy does not just have objects being present in mental states, it also has the mind being present in the world. The perceptual mental states that, on the view I draw on, essentially involve objects as their constituents, are, after all, still mental states, so also essentially involve mind. One could call ‘phenomenalist’ a view which ties the existence of physical objects to actual and possible instances of these mental (and physical) states. This would be significantly different from the versions of phenomenalism I reject, which see physical objects as constructions out of actual and possible merely mental states. If someone wants to call this ‘phenomenalism’ then I agree with Stang that my position is not different from all possible versions of phenomenalism; what is important to me is that Kant does not have the version of phenomenalism which sees the world in space and time as a construction out of actual or possible merely mental states.

Stang has two other main worries. The first concerns my saying that the things that appear to us also have a way they are in themselves. As I document in the book (chapter 1), Kant speaks like this throughout the Critique. Stang worries that this gives us too much knowledge of things in themselves, since, most obviously, it seems to give us knowledge how many things in themselves there are. I do not think this follows. First, note that at most it could give us knowledge of how many correlates of appearances there are, and not how many mind-independent things there are, since it says nothing about the number of monads, Cartesian souls and other noumena. Further, assuming a secondary quality view of colour, it seems possible both that the phenomenal colour qualities we experience are grounded in some feature of reality as it is independently of our perceiving it (and independently of its appearing coloured to us), and also that, at the level of reality as it is independently of our perceiving it, we do not find divisions that map onto our colour experience. For example, there could be more than one mind-independent property that appears to us as red. In this case, we could say both that colour is a presentation of something that has a way it is independently, and that our colour experience only allows us to group objects at the level of colour experience. Thus we can claim both that the very same reality that appears a certain way to us has an intrinsic nature that grounds these appearances, without claiming that this gives us a way of individuating reality in itself. However, unlike the merely mental version of phenomenalism, we can still do justice to Kant’s frequent talk of things as they appear to us and these same things as they are in themselves.

Stang thinks that my picture leaves open a sceptical worry, in not having an adequate account of what grounds the fact that objects have the properties they actually possess, which sometimes differ from our actual experience. As I see it, we need to keep separate the question of what is it for objects to possess some manifest quality (which essentially involves an account of how it can appear to us), and the question of what grounds the fact that objects have those qualities (which involves features of reality as it is in itself). Stang worries that, if the property that grounds objects having the manifest properties they do is not itself something manifest, then we do not have a guarantee that we could not, at the idealized limit of science, be entirely wrong. As I read the Critique, Kant’s central concern is not responding to scepticism about the external world or empirical knowledge. Giving an account of the conditions of the possibility of empirical knowledge does not require giving an account which guarantees that we could not be wrong. As is also brought out by Newton’s comments, our reading of Kant’s arguments is deeply affected by what questions we take him to be answering. There are no quick answers here, and I am happy to have been able to discuss these questions further with four such excellent interlocutors.