2. The source of the synthetic reading

It is plausible that one of the central motivations for the synthetic reading of AR comes from the very passage in the Transcendental Aesthetic in which Kant introduces the concept of an appearance. He writes:

To understand the synthetic reading of AR, we can give a plausible reconstruction of this passage, which we will later reject. It can go something like the following:Footnote ¹¹

1. (1) There are objects which affect our capacity for representations.

2. (2) These objects must be something outside the capacity for representation.

The reason for (2) is straightforward: something within our capacity for representation cannot have an effect on it, especially as to produce sensation. From here, synthetic readers will each say something like what follows:

1. (3) Appearances are the objects that affect our capacity for representations; they thus become contents of intuition.Footnote ¹²

Since it is natural to assume that a mere representation cannot itself affect us from outside, the relevant ‘objects’ might seem to be external objects or physical things. If these are identical to appearances, it follows that appearances are physical things. I’ll call this claim APT. The main difference among synthetic readers will be how they identify those objects vis-à-vis ‘things in themselves’. Some two-world readers will say:

1. (3a) The objects that affect our senses are things distinct from ‘things in themselves’.Footnote ¹³

While one-world readers will say:

1. (3b) The objects that affect our senses are the same things as those that are some way ‘in themselves’.

In either case, the key to understanding the synthetic reading is that the basic concept of appearances is that of the things that affect our senses. By any account, this makes appearances (or at least outer appearances) nearly equivalent to physical objects. The substantive differences between rival interpretations just concern the metaphysical status of mere physical objects.

Supposing, however, that synthetic readers could agree with this (very indeterminate) account of <appearance>, they should also agree that it is not an analytic or conceptual consequence of <appearance> thus defined that appearances are mere representations. For it is not an analytic claim that physical objects are representations. This could only be a substantive metaphysical thesis. Unsurprisingly, this is universally recognised among phenomenalist readers, each of whom accepts the definition of <appearance> as something like an ordinary physical object.Footnote ¹⁴ But it is equally agreed among non-phenomenalist readers that the ‘objects’ Kant speaks of in this passage are empirically real things. They tend to identify appearances with the ‘things that appear’.Footnote ¹⁵ Hence, far from admitting that this identification of appearances and representations is analytic, they will deny it, at least when ‘representation’ is taken in a straightforward sense.

Realist readers want to avoid the apparent phenomenalist metaphysics in AR and so must offer some alternative explanation for what Kant means in AR passages. Here, for example, is Allais:

One can agree with Allais that the translation of Vorstellung does not need to be ‘representation’; perhaps ‘presentation’ is a better and more neutral choice. But nothing much hangs on that, if one considers how many ways Kant formulates the fact that, without the mind, appearances would not exist at all.Footnote ¹⁶ Allais attempts to show how those passages can be accommodated, but they are clearly against the grain of her reading. Several other realist authors attempt a similar work-around with Kant’s language of representation.Footnote ¹⁷ Though I do not find any of these solutions convincing, I so far only want to show why this move seems to be necessary, given a synthetic reading of AR. For it is impossible to maintain that physical objects are representations (in a normal sense) while being committed to a physical realism about them.

Interestingly, then, both phenomenalist readers and their opponents tend to agree on the basic meaning of ‘appearance’. They only disagree about the substance of Kant’s metaphysics. I conjecture that this agreement tends to arise from an admittedly plausible reading of the passage in which Kant first introduces the term ‘appearance’.

3. The relative meaning of ‘object’

Why is the synthetic reading of AR treated as the default one, even by readers who otherwise disagree about Kant’s transcendental idealism? Pretty clearly, the explanation is Kant’s use of the term ‘object’ (Objekt, Gegenstand)Footnote ¹⁸ to describe or define <appearance>. It is often thought that Kant introduces a use of the word ‘object’ in philosophy that is close to the standard contemporary one. For example, Kant is thought to invert the uses of ‘subject’ and ‘object’ that might be found in Descartes.Footnote ¹⁹ Unlike in Descartes, in Kant ‘objects’ were no longer something necessarily in the mind. In general, modern languages tend to use ‘object’ to refer to middle-sized concrete material things: roughly, concrete particulars. Footnote ²⁰ This is not to say objects must be material things, but since that is the paradigm case, the use of ‘object’ to refer to (e.g.) numbers and other abstracta tends to generate puzzlement and philosophical controversy. Moreover, the contemporary sense of ‘object’ does not imply a connection to minds or representation. Something could be an ‘object’ (in this sense) even if it had no connection to us.

It may be true that Kant stands awkwardly on the cusp of a linguistic transformation in the use of ‘object’, one that would be in evidence in Frege and which Strawson calls ‘objects in [the] weighty sense’ (Reference Strawson1966: 73). Nicholas Stang, for example, has argued that Kant has a ‘quantificational’ sense of ‘object’, which would make it both a count noun, as in the contemporary usage, and carry with it existential commitment.Footnote ²¹ That is, in my preferred terminology, there is some usage of ‘object’ in which it is an absolute term and means as much as ‘thing’.Footnote ²² ‘Thing’ (Ding) is an absolute term with no relative usage; hence, it is always a more precise term when we want to make a claim without the suggestion of relativity. However, though there may be a use of ‘object’ in which it is nearly equivalent to ‘thing’, such usage evidently stands alongside one in which ‘object’ is explicitly or implicitly a relational term, not an absolute one that could be suitable to classify something in a stable way (i.e., so that the question, ‘Is x an object or not?’ would be a substantive one).Footnote ²³ I mean that a large number of Kantian uses of the term are embedded in contexts like ‘object of …’ (‘object of an intuition’, ‘object of experience’, etc.) or imply such a context. In these cases, to speak of something as an object(-of-x) is to say something like this: it is the target of some representation or act of cognition.Footnote ²⁴ This use of ‘object’ is a role-relative one. Just as targets in archery can be real, non-relative things outside of archery (e.g., a bale of hay), objects(-of-x) can be real, non-relative things outside of representation.Footnote ²⁵ But to say that something is a target is only to specify its role relative to archery; to say that something is an object (in the relative sense) is only to specify its role relative to representation or cognition.Footnote ²⁶ This of course does not mean that what is thus an object has an existence that is relative to representation or cognition.Footnote ²⁷ But something would not be an object in this sense apart from this role.Footnote ²⁸ This is, so far, merely a grammatical claim, not a metaphysical one.

To see that there must be a relative use of ‘object’ which is not equivalent to ‘thing’, consider a passage like the following one: ‘Yet because all representations, whether they have outer things [Dinge] for their objects [Gegenstände] or not, yet in themselves, as determinations of the mind, belong to the inner state…’ (A34/B50, emphasis added). In this passage, representations all have ‘objects’, but sometimes they are of ‘things’ and sometimes not. Clearly, we cannot substitute ‘things’/Dinge for Gegenstände here, which would say that representations can have things for their things; nor could we make the converse substitution, that they have objects for their objects. ‘Objects’ must mean something closer to targets and be conceptually connected to representation.

Notice that if and when ‘object’ is a relative term, its grammar is closer to words like ‘topic’ or ‘subject matter’ than ‘thing’. These words have an analogous relativity since a topic or subject matter belongs to that which has it as a topic or subject matter (say, a newspaper or magazine) while obviously referring to something outside as well. Further, it similarly makes little sense to ask whether something is a topic or not. But it is noteworthy that, in the context of a representational enterprise, it is not uncommon to substitute the relative term for the absolute one. Consider a pair of cynical journalists. They are following a lead to track down information about a group of suspected criminals. But, before they complete their investigation, the suspects are killed by an outside party. Now comes the morbid joke: ‘Looks like our cover story just died’. The journalists are not idealists or phenomenalists. They don’t think the suspects are their cover story full stop, but within the representational enterprise, the suspects are the cover story. Hence, the cover story meets the same fate as those people. In a similar way, we can refer to things as ‘topics’ or ‘objects’ (terms which may be conceptually connected to representation) without suggesting that the things exist in our representations of them.

Whatever the connection between the relative and absolute uses of ‘object’, the moral of the story thus far is that textual evidence regarding Kant’s use of ‘object’ will often, if not always, be inconclusive between the relative and (potentially) absolute uses. Yet, interpreters frequently put forward claims in which Kant calls appearances ‘objects’ as straightforward evidence in favour of APT.Footnote ²⁹ I hold that these texts are rarely unambiguous pieces of evidence since Kant may be using ‘object’ in a role-relative way. In any case, I will argue in the next section that the sense in which appearances are objects is the role-relative sense; if there is an absolute use of ‘object’ in which ‘object’ means ‘thing’, appearances are not objects in that sense. This makes ‘thing’ (as a translation of Ding) a more precise word for the defender of the synthetic reading of APT to make their claim (and for me to challenge), since it has no comparable grammatical ambiguity to ‘object’. The defender of APT should show that appearances are (ontologically) things.Footnote ³⁰ Later, I will argue that the relationship between appearances and things is straightforward – and mutually exclusive. Hence, passages that speak of appearances as objects are systematically ambiguous.Footnote ³¹ If there is a use of ‘object’ that means ‘thing’, we can use ‘thing’ to paraphrase those passages. If there is another relative use of an object, which is in itself entirely neutral vis-à-vis ontology, we can disambiguate this explicitly. Henceforth, when I assume that ‘object’ should be understood in the relative sense, I will add an ‘r-’ prefix.Footnote ³²

4. The definition of appearance

We have so far established that Kant’s use of ‘object’ is often relative, and so we cannot conclude from his description of appearances as objects that appearances are things. However, there is still further clarification needed once the relative use of ‘object’ has been granted. Recall that the starting piece of textual evidence that apparently favours APT is the definitional text from the Transcendental Aesthetic. Though that passage does contain an instance of ‘object’ that seems to be an absolute one (the ‘effect of an object on the capacity for representation’),Footnote ³³ the use of ‘object’ that falls within the definition of <appearance> seems to be a patently role-relative one: ‘The undeterminedFootnote ³⁴ object of an intuition is called appearance [Erscheinung]’ (A19–20/B34, underlining added). This passage seems to present us with a definition,Footnote ³⁵ and since I am claiming that AR is analytic, or true by definition, it is my aim to draw out AR from this definition. The definition is:

$$\boldsymbol {\textrm {Appearance}}{ \,=\, _{{\rm{def}}}}\,{\rm{the\, r-object\, of\, an\, empirical\, intuition}}$$

But this definition can be disambiguated in two ways, each of which respects the lessons learned about ‘object’ in the previous section. A synthetic reader can acknowledge the frequent occurrence of the r-object and treat the definition as meaning:

$$\boldsymbol {\textrm {Appearance (realist)}}{ \,=\, _{{\rm{def}}}}{\rm\,{the\, thing\, which\, is\, the\, r-object\, of\, an\, empirical\, intuition}}$$

This definition takes into account the lessons learned above because, just as the target of archery can be a bale of hay, an r-object of intuition can be the thing that is supposed to be represented by the intuition. However, the analytic reading of AR requires the definition to be understood as follows:

$$\boldsymbol {\textrm {Appearance (representationalist)}}{ \,=\, _{{\rm{def}}}}\,{\rm{the\, r-object\, presented\, in\, an\, empirical\, intuition}}$$

Briefly, the realist definition makes AR synthetic because it makes the genus of appearances <thing>, and it is not analytic that things are representations. The representationalist definition makes AR analytic because here ‘r-object’ is simply a term for what is in a representation. The analytic reading simply requires some sense of ‘r-object’ that makes it belong essentially to representation. This sense, of course, is akin to the (Brentanian) notion of an intentional object.Footnote ³⁶

It is worth distinguishing the two understandings of r-objects with an analogy, which, though it risks oversimplification, helps to illustrate this difference in stark terms. Suppose you are watching a documentary about JFK. If asked about what is presented in the film, it is natural to respond that it is simply JFK: JFK is the film’s r-object. But this can mean two things. First, it could mean that JFK himself (the now-deceased US President) is presented in the film. But that is potentially misleading. Only images of him appear, and real people, even when they are living, do not exist in the screens that present their images. The second option is that the film presents a JFK rendition that is meant to stand for that deceased person. This rendition of JFK exists in the images, sound, and text that are actually available to the film’s audience. The same would be true of a fictional film portraying someone who never existed, say Frodo Baggins. The Lord of the Rings films present a rendition of Frodo to their audience, even though there is nothing else this version stands for, outside of their own minds.

I need not belabour the distinction, for it is rather obvious. But it is important to make clear that the realist definition of <appearance> makes it out so that an appearance is a thing that has become the object of a representation. In the film analogy, that makes appearances analogous to the following description: the actual US president who has become the subject matter of a film. This conception still makes appearances r-objects of representations, but not representations themselves. On that view, appearances are things on their own account, but they become representations (or perhaps possible representations) through a relation to our intuition. But appearances on the representationalist reading are more akin, in the film analogy, to the actual images presented to the viewer of the film, rather than to any person. Let us say that the real JFK, or the person depicted in any documentary, is the distal object of the film.Footnote ³⁷ By contrast, the JFK rendition that is constituted by the materials of filmmaking and presented to audiences is the immanent object of the film. For Kant, appearances are r-objects immanent in the contents presented in empirical intuitions.

Though the word ‘object’ (even taken in a relative sense) can be used for the distal object and for the immanent content of a representation, this does not mean that Kant’s definition of <appearance> is ambiguous, for Kant’s definition of appearance and his consistent use of the term are concerned only with the immanent content of the presentation, something that is intrinsically presentational.Footnote ³⁸ Some will grate against the idea that ‘object’ for Kant can mean something completely immanent to presentations rather than their distal correlate. But Kant frequently speaks this way.Footnote ³⁹ He says, ‘For as appearances they amount to an object which is merely in us, because a mere modification of our sensibility cannot be encountered outside of us at all’ (A 129). This passage presents appearances as if they exist in the very ‘fabric’ of sensibility, much as a JFK rendition exists in the materials of film. He also writes:

In this passage, if we follow the connecting pronoun ‘which’, not every representation is an r-object, but some representations present r-objects, which (for Kant) makes those r-objects representations as well. To be sure, Kant is not identifying all representations as r-objects in the relevant sense. Here, the representations that are r-objects are also connected with the laws of the unity of experience. But he does not identify the r-objects of experience with distal things but with representations with which we are affected. He makes a related point in the Second Analogy:

Here, appearances are ‘as representations’ r-objects: I take this to mean that they are empirical contents presented in representation. But they also ‘designate’ a further (distal) object: their referential target is something outside the representation.Footnote ⁴⁰ This is perhaps why ‘appearance’ also implies something (else) that appears (cf. B xxvi–xxvii; A249).

Now, in the case of a film, it is easy to distinguish the immanent r-object (i.e., the lively JFK rendition of the documentary) from the distal r-object (i.e., the deceased JFK). But appearances are empirical representations, and moreover, they are perceptual representations.Footnote ⁴¹ The unavoidable difficulty with this account in perceptual contexts is that the presented r-object and the designated distal (or ‘transcendental’) object seem to the perceiver to coincide: actual perceptual representations announce their distal objects to us, but they do not typically announce themselves as representations.Footnote ⁴² If I am (foolishly) looking at the Sun, it is natural to say that the Sun is the r-object of my intuition. But, if we follow standard phenomenalist accounts, to say that the Sun is my r-object involves a great simplification. What I call perceiving or intuiting the Sun means that there is for me a Sun-shaped image in my field of view. The appearance is precisely that immanent rendition of the Sun contained in my sensory intuition. But the Sun itself – the distal object, or the thing – is something that exists apart from this immanent rendition. At least upon reflection, I take the Sun-as-it-appears-to-me (a content in my empirical intuition) to ‘stand for’ a Sun that would exist outside all my representations. This is what Kant seems to have in mind when he alludes to the Critique’s teaching us ‘to take the object in a twofold significance, namely, as an appearance [sc. as immanent to the representation] and as a thing in its own right [sc. as what is ultimately signified, i.e., the transcendental object]’ (Bxxvii). This ‘twofold significance’ does not seem to be a special ambiguity with Kant’s use of the term ‘object’, but rather something bound up with representation as such: taking something as a representation implies taking it to stand for something. What is important, on my account, is that though appearances both are an r-object and designate an object, the r-object that they are is purely immanent to representation.

Thus, when Kant says, for example, that a house is an appearance (A190/B236), he can only mean – and I think in the context he does mean – a house that is currently being targeted (or pretended to be, for the sake of an example) in an empirical intuition (= ‘which stands before me’ [das vor mich steht]) and for that reason is the immanent content of a perceptual representation. That house (the distal object) is a real thing outside the intuition, but, since our access to it is only via a perceptual representation of the house – imbued with colour, touch, and perspective – we must acknowledge that the ‘house’ available to us is a represented house, and thus that it is a representation. This does not mean that (distal) houses are mere representations.Footnote ⁴³ For part of what it means to take the house-representation as an appearance is precisely to see it designating something that does not appear otherwise than in the representation.Footnote ⁴⁴

The present account overlaps in significant ways with Tolley’s (Reference Tolley, Schafer and Stang2022) recent ‘Fregean’ interpretation of appearances. Tolley argues that appearances should be understood on the model of Fregean object-dependent senses. On this model, an indexical expression like ‘this house’ has a distal referent, presumably a particular house. But the expression has a publicly identifiable meaning – its sense – that can be understood by anyone who uses the expression correctly. Though the senses of these expressions are not identical to objects, following Fregeans like McDowell and Evans, Tolley notes that the sense of such indexicals is object-dependent in the following way: we could not understand the expression as having the meaning it has except insofar as its referent exists and is present to the mind (Tolley Reference Tolley, Schafer and Stang2022: 217). The reference is not included in the sense (à la Russell), but the sense could not be the sense it is apart from the actual (distal) object referred to. Moreover, everyone who understands the expression understands precisely that ‘one’ public sense.

Tolley understands appearances as object-dependent senses in this latter way. This permits him, like the present account, to preserve the way in which appearances are akin to images and not identical to the objects they (distally) designate (Tolley Reference Tolley, Schafer and Stang2022: 225–6). However, just as Frege introduces the notion of sense to help de-psychologise the concept of meaning, insisting as he does that senses are not private mental representations (precisely Kant’s Vorstellungen), Tolley argues that Kantian appearances are similarly public and not mental, constituting something like a more concrete version of Frege’s ‘third realm’ (Reference Tolley, Schafer and Stang2022: 234–5).

Thus, if the Fregean could acknowledge that AR is analytic, he differs on the meaning of ‘representation’, for an appearance would no longer be a mental representation.Footnote ⁴⁵ Tolley, along with many others, seems to think that it is a desideratum of an interpretation of appearances that they be public entities by some description. By contrast, the above account makes appearances decidedly private. To speak crudely, my own appearances are in my own mind, and yours are in yours. Recall:

We take these appearances to ‘designate’ common distal objects of experience, but this does not alter the merely mental existence of the appearances.

Now, is it a desideratum that appearances are not private in this way? It is clear that just about everyone will have to acknowledge that there is something mentally private for Kant. If not appearances, certainly our empirical intuitions (the acts, at the very least) are private to each of us. But I am claiming merely that Kant uses ‘appearance’ to refer to contents bound up with private acts of empirical intuition.Footnote ⁴⁷ Unless a view aims to deny that there are such private mental contents,Footnote ⁴⁸ everyone will need a concept for them, whether they are phenomenalists or not. In my view, <appearance> is Kant’s concept for them.

It is only problematic that appearances are private, I believe, if we accept the synthetic reading of AR. For then appearances are first understood as ‘objects’ (read as ‘things’), and it is imperative that (physical) things are not merely private mental entities. But if appearances are not supposed to be things, it is not a problem that they are not public. That is, so long as we have some other concept to describe the entities that are supposed to be public in that way. Now, one might think that, without the concept of appearance, the important role for ordinary things is left vacant in Kant’s account (since by common agreement it cannot be played by ‘things in themselves’). I will address this concern in the following section. If it can be addressed adequately, perhaps the main obstacle to acknowledging AR as analytic can fall away.

5. Realism or phenomenalism?

Though on the reading I have offered, AR is analytic and does not present a metaphysical view at all, there is a more traditional reason for supposing that AR leads to a version of metaphysical phenomenalism or subjective idealism and hence against reading it as merely analytic. For if ‘appearance’ is a term wholly reserved for representational contents, then we seem to have no term left for the distal thing other than the ‘thing in itself’.Footnote ⁴⁹ And this looks like a problem because ‘things in themselves’ are supposed to be non-spatio-temporal entities, or things which have only non-appearing properties, or something like that. It is hard to see how we could say that the perception of a cube is of a distal cube outside the representation if nothing outside representations is cubical at all.Footnote ⁵⁰ If phenomenalists claim that the ‘thing in itself’ plays some causal or affectional role with respect to the appearances, this must be a very indirect role, because that ‘cube’ won’t be cubical (not to mention the ancient problem of attributing causality to it). But it might be said that this is at least a consistent option. Realists, by contrast, have tended to think that they are required to identify the ‘thing in itself’ as the same thing as the appearance. This creates notorious difficulties about how one and the same thing could share properties that are taken to be in conflict.

Should we simply concede metaphysical phenomenalism? To be sure, the above view is compatible with a stronger version of phenomenalism than I endorse, though not without raising new difficulties for it. For, as I suggested above, metaphysical phenomenalists typically endorse APT: appearances are physical things. A phenomenalist who also accepts the analytic reading of AR must reject APT, considered as a definition. For APT cannot be analytic if AR is (it cannot be analytically true that things are representations), and, as we saw above, it is the conjunction of APT with the synthetic reading of AR that typically gets phenomenalism off the ground. The view available to metaphysical phenomenalists, in place of APT, is apparently to deny the existence of physical things that are not representations. This is a step I cannot take and one that I think Kant did not.

It is worth exploring whether a realist interpretation of Kant can better embrace the analytic reading of AR. I believe it can. What must be shown is that Kant leaves room for a kind of entity that is on the one hand not an appearance (since these are only representations) but could have the properties that appearances present their distal objects as having: shape and size, for example. The latter requirement is what must be the case for our talk of physical things to have anything like a standard meaning.

In recent work (Wolf Reference Wolf2023, Reference Wolf2026), I have tried to show how mere things (Dinge) can satisfy these demands. Whereas it is commonly thought that ‘thing’ in Kant is ambiguous between ‘thing in itself’ and ‘appearance’, I argue that Kant’s use of the term Ding is basic and fundamental; it does not require a disambiguation between ‘appearance’ and ‘thing in itself’. Dinge are the mind-independent units of reality, and we do not need the expression ‘in itself’ to specify their mind-independence. On this view, if Kant describes something as a ‘thing that appears’ or a ‘thing that affects our sensibility’, it is a thing, full stop, and (by assumption) mind-independently real. Yet, in contrast to other views, we should not equate ‘thing that appears’ with ‘appearance’, or say that appearances are things in some aspect, since appearances are solely on the representational side of the divide. And mere things do not have to be or contain appearances to be represented by them.

This proposal is well illustrated by the following passage from the Prolegomena:

In this passage, mere Dinge are given to us as objects of our senses. They are the distal targets of our representations. We are tempted to ask: are these things appearances or ‘things in themselves’? But we should resist this question, for ‘things’ is a sufficient answer to the question ‘What exists outside of us?’ As other one-world interpretations imply, Kant simply tells us that we do not know what these things are like ‘as they may be in themselves’. For now, we can set aside what kind of ignorance this entails.

The present interpretation of the Prolegomena passage is quite consonant with a moderate metaphysical reading of Kant.Footnote ⁵¹ The only modification that must be pressed is what the passage says about appearances, which is just what the analytic reading of AR requires: they are ‘the representations that [things] produce in us because they affect our senses’. Since appearances are produced in us by things, appearances are not identical to things. A thing does not have any ‘aspect’ that is an appearance.Footnote ⁵²

Yet, we are still left wondering what a (mere) Ding could be like on a Kantian view. In particular, we may well worry about the spatio-temporality of mere Dinge. For suppose appearances represent things as being spatio-temporal, but nothing outside of representations is spatio-temporal. Since I can make no sense of the coffee cup in front of me if it has no spatio-temporal properties, any view that says there is no spatio-temporal coffee cup might as well say that the coffee cup is a mere ‘virtual object’,Footnote ⁵³ if it exists at all. Phenomenalism strikes back.

However, I believe that when we recognise <thing> as a distinct and unambiguous concept, we are also led to modify our conception of what it means to deny knowledge of things ‘in themselves’. The second part of my proposal is that the expression ‘in itself’ (an sich selbst) is often better translated as ‘per se’ and that a grammatical analysis shows that it does not combine with nouns to create more complex nouns.Footnote ⁵⁴ The expression must be understood in the context of some predicative attribution. For example, attributing a property to an F per se is attributing it to F without qualification, and so in virtue of being a mere F. In this way, ‘per se’ (or in itself) is syntactically comparable to modal operators like ‘possibly’ and ‘necessarily’. This implies that the collocation ‘thing in itself’ is not a noun phrase, since it includes a predicative adverb. As a result, it cannot designate a kind of thing. Instead, statements that use ‘things in themselves’ can be analysed into statements about mere things, without being equivalent to them.

From this analysis, it follows that claims that are supposed to be true when we use ‘thing in itself’ are not obviously true when we substitute ‘thing’. If it is true that ‘things in themselves’ are not in space (or ‘things are not per se in space’), is it also true that things (with no modification) are not spatial?Footnote ⁵⁵ This inference does not follow immediately. For compare the claims:

1. (1) Guns are not dangerous per se

2. (2) Guns are not dangerous

(1) might well be true, but (2) is not, and it does not follow from (1). Just so, the interpretative tradition has focused on the analogue of (1) for the spatiality of things. And since the unmodified concept <thing> has received little attention in the literature,Footnote ⁵⁶ the question about the spatiality of mere things (the analogue of [2]) has hardly been asked. I have argued elsewhere (Wolf Reference Wolf2026) that if we consider what Kant’s views imply for mere Dinge, his arguments in the Transcendental Aesthetic do not mean to establish that no Dinge have properties that we call spatial, such as size and shape; they establish the weaker claim that space, since it belongs essentially to human sensory representation, does not belong to the nature of things as such, what things are qua things.Footnote ⁵⁷ If Kant’s denial about what things are per se is not directly a denial that some things are spatial,Footnote ⁵⁸ there is logical room in Kant’s view for Dinge that have the characteristics they appear to have. So long as things of some such kind are available, we can grant Kant a robust picture of the contents of the mind without detracting from a robust account of reality outside the mind. Appearances can be wholly representational, because we can permit the existence of things that are (at least possibly) veridically represented by the appearances.

To be sure, I have not said enough to make my account of mere things wholly convincing, and further alternatives are worth exploring. But I believe the analytic reading of AR makes available new interpretative possibilities, as it enables a greater logical independence between <appearance> and <thing> than has typically been appreciated.

To take a significant example, in light of our proposal, it is illuminating to examine a passage that has been puzzling in itself, but especially so on a realist construal of appearances. The passage is one in which Kant discusses secondary qualities in the Prolegomena. He writes:

This passage is supposed to convince Kant’s early readers that he is not a Berkeleyan idealist, but, on the surface, it might look to be proof positive that he is one.Footnote ⁵⁹ At first glance, it may look as if Kant is arguing something like this:

Thus conceived, the argument is doomed. For it is hard to see what could be more idealist than claiming that all the properties of bodies are mind-dependent. It was precisely the ‘some’ that made it plausible that Lockeans were hanging on to realism in some form. ‘All’ makes all the difference.

Yet, that is not Kant’s argument. Instead, in my view, he is making a simple point about the relation between representation and mind-independent reality. It is something like this: When Lockeans attribute colour to ‘appearances’ or to the content of empirical representations, they do not detract from the reality of the things that appear. Lockeans are not idealists, because this addition to our representations is not necessarily a subtraction from the things (after all, Lockeans have to posit mind-independent dispositions in things to account for so-called secondary qualities). Accordingly, I, Kant, claim that representation by itself contains more than even a Lockean would countenance: the a priori forms of space and time belong to representation whether or not they belong to things. Still, in the same way, by adding this content to appearances, I am subtracting nothing from the things that are thus represented. Hence, I cannot be an idealist.

On this reading, Kant is an epistemic phenomenalist: he believes that we have no cognition of anything apart from the sensory conditions of our representations. As he writes in the Second Analogy, ‘[W]e have only to do with our representations; how things in themselves may be (without regard to representations through which they affect us) is entirely beyond our cognitive sphere’ (A190/B235; cf. A108–9). This is why Kant is sometimes willing to say, for example, that the bodies we perceive are only representations: because the properties we can perceive them to have depend wholly on our representational powers. Yet, in the Prolegomena passage, Kant rightly recognises that this epistemic phenomenalism is itself simply innocent of metaphysical theses.Footnote ⁶⁰ A more complete epistemic phenomenalism than Locke’s says nothing against an equally complete physical realism. Hence, this passage relies on the representationality of appearances or the contents of sensible intuitions, precisely to demonstrate Kant’s realism about the things that appear.