SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

According to the ‘One Object’ reading of Kant's transcendental idealism, the distinction between the appearance and the thing in itself is not a distinction between two objects, but between two ways of considering one and the same object. On the ‘Metaphysical’ version of the One Object reading, it is a distinction between two kinds of properties possessed by one and the same object. Consequently, the Metaphysical One Object view holds that a given appearance, an empirical object, is numerically identical to the thing in itself that appears as that object. I raise various indiscernibility arguments against that view; because an appearance has different spatiotemporal and modal properties than a thing in itself, no appearance can be identical to a thing in itself. I point out that these arguments are similar to arguments against Monism, the view that material objects are numerically identical to the matter of which they are made. I outline some strategies Monists have developed to respond to these indiscernibility arguments and then develop parallel responses on behalf of the Metaphysical One Object view. However, I then raise another indiscernibility argument, to which, I argue, the Metaphysical One Object view cannot respond, even using the resources I have developed thus far. I develop a modified version of the Metaphysical One Object view that can respond to this new indiscernibility argument, but, I argue, this modified version of the One Object view is only a terminological variant of the Two Object view. When the Metaphysical One Object view is fully thought through it becomes the Two Object view. I conclude that Kantian appearances are not numerically identical to the things in themselves that appear to us.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

Kant famously held that non-spatiotemporal things in themselves appear to us as the spatiotemporal objects we experience. Readers of Kant have long been divided over how to understand the relation between appearances and things in themselves. The primary division is between ‘Two Object’ readings and ‘One Object’ readings, although some readings do not fit neatly into either camp.2

According to the Two Object reading—as its name suggests—appearances are (in general) numerically distinct from things in themselves. Things in themselves are objects whose existence and nature does not depend upon how (or whether) they are experienced;3 these objects appear to human cognitive subjects as a distinct plurality of spatiotemporal objects, appearances, whose nature and, arguably, existence, does depend upon how we experience them.4 Objects in the second category, appearances, are not (in general) identical to objects in the first category, things in themselves. I add the qualification ‘in general’ because some Two Object readers allow that in the case of the self, the appearance—or ‘phenomenal self’—is numerically identical to the thing in itself—the ‘noumenal self’. How to ‘count’ ‘selves’ on the Two Object view will not be an issue in this paper, because my focus will be the One Object reading.5

On the One Object reading, a given appearance is numerically identical to the thing in itself that appears to us as that object. The distinction between appearance and thing in itself is not a distinction between objects but between ways of considering one and the same object. One Object readings are not committed to claiming that every thing in itself appears to us, so they are not committed to claiming that every thing in itself is identical to some appearance; there may be objects that can be considered as things in themselves, but cannot be considered as appearances, although such objects are unknowable by us through theoretical means (because they do not appear to us). But they are committed to claiming that every appearance is numerically identical to a thing in itself, since every object that can be considered as an appearance can also be considered as a thing in itself.

There are as many different versions of the One Object reading as there are different ways of distinguishing between what it is to consider one and the same object ‘as an appearance’ and ‘as a thing in itself.’ The primary distinction among One Object readings is between ‘Epistemological’ and ‘Metaphysical’ readings. On the Epistemological reading, the distinction between the two ways of considering the object is not a metaphysically robust distinction among the properties of the object under consideration but a distinction between two different perspectives we can take on it: the empirical and the transcendental.6 To consider an object from the empirical perspective is to consider the object under the sensible conditions which make possible our knowledge of it, space and time; to consider the object from the transcendental perspective is to abstract from these sensible conditions and consider the object merely as an object of knowledge in general. Thus, on the epistemic reading, if space and time are sensible conditions of the possibility of our knowledge of objects, it follows immediately that the thing in itself is not determinately spatial or temporal, because in considering an object as a thing in itself we abstract from sensible epistemic conditions.7 The only properties we can ascribe to an object when considering it as a thing in itself are the properties that follow from the very idea of being the object of a finite knowing subject; these may include existing independently of us and causally affecting us but they do not include being spatial, because there may be beings with non-spatial sense perception (intuition).

Metaphysical One Object readings interpret the appearance/thing in itself distinction as a distinction between two classes of properties possessed by one and the same object; call them the ‘phenomenal’ properties and the ‘noumenal’ properties. The appearance is the object qua bearer of phenomenal properties and the thing in itself is the object qua bearer of noumenal properties.8 Conceptions of this distinction that have been defended in the literature include: properties the object appears to have versus properties the object in fact has; properties the object has relative to us versus properties the object has ‘in itself’; relational properties of the object versus intrinsic properties; response-dependent properties versus properties the object has independently of the responses of subjects, etc.9 In this paper I'm going to abstract from these differences and discuss Metaphysical One Object readings as such. In fact, Metaphysical One Object readings have more in common with Two Object readings because each grant a certain kind of metaphysical ‘reality’ to the thing in itself: either as a distinct object from the appearance, or as the bearer of properties in itself and independently of us. Both readings understand Kant's doctrine of our ignorance of things in themselves as meaning that there are significant features of reality of which we are necessarily ignorant. Furthermore, both readings take literally Kant's numerous claims that things in themselves are the ‘grounds’ of appearances and that the noumenal world of things in themselves is more fundamental than the world of empirical objects. By contrast, Epistemological readers eschew any talk of how the object is in itself and independently of us as a confused bit of precisely the kind of metaphysics Kant was trying to overcome. They typically adopt deflationary interpretations of Kant's doctrine of our ignorance of things in themselves, according to which the point of this doctrine is not to deny us knowledge of some substantive domain of facts about reality, but merely to remind us that we can only know objects under certain conditions, and that the idea of knowing an object independently of the very conditions which makes knowledge possible is incoherent.10 Finally, Epistemological readers deny that either standpoint, the empirical or the transcendental, is more fundamental or better reveals how things ‘really are’; they maintain, instead, that the empirical standpoint is appropriate in certain contexts (e.g. the context of empirical natural science) while the transcendental perspective is appropriate in other contexts (e.g. transcendental philosophy, especially practical philosophy).

Perhaps interpretations of Kant's idealism should be divided, in the first place, into ‘Metaphysical’ and ‘Epistemological’ readings, and then the former should be further sub-divided into ‘one object’ and ‘two object’ views. I have not done this here because I would find it difficult to explain what the Epistemological reading is other than by contrasting it specifically with the Metaphysical One Object view.

Although I think Epistemological One Object readings are untenable, my argument against them lies outside the scope of this paper.11 In this paper I'm going to focus on Metaphysical One Object readings, so I will drop the qualifier ‘Metaphysical.’ In the first part of the paper I'm going to raise several problems for the One Object reading by applying ‘Leibniz's Law’: identicals are indiscernible.12 (This is not be confused with the famous Leibnizian principle of the identity of indiscernibles, the converse of Leibniz's Law.) The reason Leibniz's Law creates problems for the One Object reading is that Kant appears to ascribe different properties to appearances and to things in themselves, which violates Leibniz's Law if the appearance and the thing in itself are one and the same object.13 Appearances and things in themselves differ in spatiotemporal properties and modal properties. Finally, I will argue that the One Object reading raises problems about individuation: if appearances and things in themselves are numerically identical and we know them to be numerically identical then in individuating empirical objects we obtain knowledge of the individuation of things in themselves, which appears to violate Kant's restriction of theoretical knowledge to the empirical world.

In the rest of the paper I show how the One Object reading can surmount these problems by comparing it to a similar view in contemporary metaphysics, the Monist view of material constitution. Take some ordinary material object and the matter of which it is made, for instance, a sword and the steel from which it is forged. Are they one object or two? In other words, are material objects numerically identical to the matter of which they are made? The Monist answers ‘yes.’ But the Monist faces several well-known problems, since material objects seem to differ in various ways from the matter of which they are made. For instance, the sword has different ‘sortal’ properties: the sword, after all, is a weapon, while the metal is not. The sword has different ‘temporal’ properties: the metal existed before the sword was forged, and will continue to exist after the sword is beaten into a plowshare. Finally, the sword has different modal properties: the metal would continue to exist if it were beaten into a plowshare but, intuitively, the sword would not.

Monists about material objects have developed strategies to answer these objections. I explain what Monism is and how it solves these problems in section two. In section three I show that the problems I raised in section one for the One Object view are analogous to the problems with the Monist view, and then I show how the One Object reader can apply the same strategies developed by Monists to block these objections. In section four I raise a new indiscernibility argument against the One Object view, the synchronic indiscernibility argument. I show that the One Object responses to the earlier indiscernibility arguments, modeled on Monist strategies for resisting similar arguments, cannot be used to respond to the synchronic indiscernibility argument. The One Object reader needs something different. I then show how the One Object reader can modify her view to respond to the synchronic indiscernibility argument. I argue that this modified version of the One Object view is only terminologically different than the Two Object view. There is no substantive difference between the One Object view, when it is modified to respond to the synchronic indiscernibility argument, and the Two Object view. Either the One Object view has to be formulated in a way that makes it subject to that indiscernibility argument, or it must transform into the Two Object view.

There is a widespread sense among Kant scholars that the question whether the appearance and the thing in itself are one object or two ends in a ‘textual stalemate’: there are passages that favor a One Object reading, and passages that favor a Two Object reading, and no obvious way of discerning which represents Kant's considered position. Thus it might be surprising that my argument will not be heavily textual; I only rely on particular texts in motivating the ‘modal indiscernibility’ argument in section one. My aim in this paper is to ‘short-circuit’ the textual stalemate by showing that the One Object view is incompatible with other very high-level Kantian commitments, especially the doctrine of our ignorance of things in themselves. In other words, my claim in this paper is that when the Metaphysical One Object is fully thought through, it becomes either incompatible with Kantian strictures on our knowledge of things in themselves, or collapses into the Two Object reading. Thus, there aren't really two positions to choose from, after all. A fully developed and thought-through Metaphysical One Object reading becomes a Two Object reading. A full and complete answer to the question ‘did Kant hold a One Object or a Two Object view?’ must weigh fidelity to the texts with philosophical coherence; we do not want to attribute to Kant a view that flies in the face of what he wrote, or that is internally incoherent. However, if I am right, this makes it very likely that the ultimate balance of reasons will tip in favor of the Two Object view: if the One Object view is incompatible with Kant's doctrine of ‘noumenal ignorance,’ one of the pillars of the theoretical philosophy, then only overwhelming textual evidence to the contrary will tip the ultimate balance of reasons back in favor of the One Object view. In advance of a comprehensive survey of the textual evidence, which lies outside the scope of this paper, I can claim only to have made the world significantly less safe for the One Object view.14

1. Problems for the ‘One Object’ Reading

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

(a) The simple indiscernibility argument

In this section I consider some problems for the One Object reading, problems that, although they have not been widely discussed in the literature, generate powerful arguments against that view. Many of these arguments share a common form. They appeal to Leibniz's Law—the principle that identical are indiscernible—and premises to the effect that appearances and things in themselves differ in certain kinds of properties. Their common form can thus be represented:

  1. The thing in itself that appears as the ϕ-appearance is not F.
  2. The ϕ-appearance is F.
  • (LL) (x)(y)(x = y ⊃ Fx [LEFT RIGHT ARROW] Fy)
  • (3)
    ∴The ϕ-appearance ≠ the thing in itself that appears as the ϕ-appearance.

where ‘ϕ’ expresses some property that only one appearance has, so that that the definite description ‘the ϕ-appearance’ refers.15 For instance, if ϕ expresses the property of being a computer on my desk, then (1) says that the computer on my desk is F while (2) says that the thing in itself that appears as that computer is not F. Since Leibniz's Law is a common premise in these arguments, henceforth I will not make it explicit. Since Leibniz's Law is sometimes referred to as the ‘Indiscernibility of Identicals’ I will refer to them as ‘indiscernibility’ arguments.The first indiscernibility argument I want to consider admits of a very straightforward and effective response by the One Object reader. However, the structure of that response, and why it is ineffective against other indiscernibility arguments, will be instructive for our later discussion. The simple indiscernibility argument is:

  1. The thing in itself that appears as the ϕ-appearance is not spatial.
  2. The ϕ-appearance is spatial.
  3. The ϕ-appearance ≠ the thing in itself that appears as the ϕ-appearance.

where ‘ϕ’ expresses some property that only one appearance has. Let t be the object referred to by the definite description ‘thing in itself that appears as the ϕ-appearance’ and a be the object referred to by the definite description ‘the ϕ-appearance.’ The point of this argument is to bring out that the truth conditions of (1) and (2) cannot, on pain of entailing (3), be given by:

  • (1*)
    t is not spatial.
  • (2*)
    a is spatial.

To put this contemporary terms, the One Object reading needs to hold that the definite descriptions ‘the thing in itself that appears as the ϕ’ and ‘the ϕ-appearance’ do not contribute merely their referent to the truth-conditions of judgments like (1) and (2) in which they appear. Different One Object readings will give different accounts of the truth-conditions of (1) and (2); for our purposes, it suffices to note that a standard such account would give:

  • (1†)
    t is not spatial
  • (2†)
    a appears spatial

which clearly do not entail,

  • (4)
    ta.

The lesson of this is that the One Object view can resist the simple indiscernibility argument by giving truth-conditions for (1) and (2) other than (1*) and (2*). The simplest such view would be that the definite description ‘the ϕ-appearance’ introduces not merely the object into the truth-condition of judgments involving it, but also modifies the predicate, so that in general judgments of the form

  • (5)
    The ϕ-appearance is F.

have truth-conditions of the form

  • (6)
    a appears F

where a is the object referred to by ‘the ϕ-appearance.’ Different One Object readings are possible. For instance, one might hold that the definite description ‘the thing in itself that appears as the ϕ-appearance’ also does not merely contribute its referent to the truth-conditions of judgments involving it, but also modifies the predicate. For instance, one might hold that judgments of the form,

  • (7)
    The thing in itself that appears as the ϕ-appearance is F

have truth-conditions of the form

  • (8)
    F is an intrinsic property of t.16

where t is the referent of the definite description ‘the thing in itself that appears as the ϕ-appearance’ in (7). Or, one might adopt a simpler model on which judgments of the form (6) have truth-conditions of the form

  • (9)
    t is F.

For our purposes, all that matters is that the One Object reader can escape from the simplest indiscernibility argument by introducing some complexity into the predicates of judgments about appearances.

So far I have considered these arguments as involving the definite descriptions ‘the ϕ-appearance’ and ‘the thing in itself that appears as the ϕ-appearance.’ I have done so because debates about Kant's transcendental idealism are often characterized as debates about whether ‘the appearance’ and ‘the thing in itself’ are identical or distinct, where these are not taken to refer to generic entities, but are taken to refer to particular appearances and particular things in themselves in particular cases (hence the need for the definite description the ‘the ϕ-appearance’). However, it might be thought that these debates go very differently if we use names, or singular terms, rather than definite descriptions. Let a be the object referred to by ‘the ϕ-appearance’ and let t be the referent of ‘the thing in itself that appears as the ϕ-appearance.’ Consider the following indiscernibility argument:

  1. t is not spatial.
  2. a is spatial.
  3. ta.

It appears that this argument is more of a problem for the One Object reader, because (1) and (2) directly predicate distinct properties of (what the One Object reader is committed to claiming are) one and the same object. The One Object reader escaped the first indiscernibility argument by claiming that the definite description ‘the appearance’ contributes not only its referent to the truth-conditions of judgments involving it, but also modifies the predicate of such judgments. But this move does not appear available here, since (2) appears to simply state of that object that it is spatial and (1) appears to simply state of that very same object that it is not spatial.17

But remember how these names were introduced: ‘a’ was introduced to name the referent of the definite description ‘the ϕ-appearance’ while ‘t’ was introduced to name the referent of the definite description ‘the thing in itself that appears as the ϕ-appearance.’ The One Object reader, therefore, has a natural reply to this modified version of the simple indiscernibility argument. Names for objects either have to be introduced empirically—through reference-fixing empirical definite descriptions like ‘the body that I am currently perceiving’ or an empirical complex demonstrative like ‘that body’—or non-empirically—through reference-fixing non-empirical definite descriptions like ‘the thing in itself appearing to me as this body now.’18 This duality of means for introducing names derives from a duality in ways of considering objects: empirically and non-empirically. Furthermore, names introduced through empirical means contribute not merely their referents to the truth-conditions of judgments involving them; they also modify the predicate. A version of such a One Object view would be that judgments of the form

  • (2)
    a is F

have truth-conditions of the form

  • (2*)
    o appears F

where ‘a’ is an empirical name and refers to object o. Names for objects thus bear the stamp of the means by which they were introduced to name those objects. In this fashion, the One Object view can escape the simple indiscernibility argument when it is formulated with names of objects rather than definite descriptions, because

  1. t is not spatial, and
  • (2*)
    o appears spatial

do not entail

  • (3)
    to.

Enough with simple indiscernibility arguments. Now I want to consider indiscernibility arguments that pose more serious challenges to the One Object reading.

(b) The modal indiscernibility argument

In a number of texts, Kant claims that empirical objects would not exist if appropriately constituted cognitive subjects did not exist to experience them:

We have therefore wanted to say that all our intuition is nothing but the representation of appearance; that the things that we intuit are not in themselves what we intuit them to be, nor are their relations so constituted in themselves as they appear to us; and that if we remove our own subject or even only the subjective constitution of the senses in general, then all the constitution, all relations of objects in space and time, indeed space and time themselves would disappear, and as appearances they cannot exist in themselves, but only in us. (A42/B59)

Why do we have need of a doctrine of the soul grounded merely on pure rational principles? Without doubt chiefly with the intent of securing our thinking Self from the danger of materialism. But this is achieved by the rational concept of our thinking Self that we have given. For according to it, so little fear remains that if one took matter away then all thinking and even the existence of thinking beings would be abolished, that it rather shows clearly that if I were to take away the thinking subject, the whole corporeal world would have to disappear, as this is nothing but the appearance in the sensibility of our subject and one mode of its representations. (A383)

One must note well this paradoxical but correct proposition, that nothing is in space except what is represented in it. For space itself is nothing other than representation; consequently, what is in it must be contained in representation, and nothing at all is in space except as it is really represented in it. A proposition which must of course sound peculiar is that a thing exists only in the representation of it; but it loses its offensive character here, because the things with which we have to do are not things in themselves but only appearances, i.e., representations. (A374n) 19

In the first two passages, Kant claims that if you ‘take away’ a cognitive subject with our spatiotemporal form of intuition, there would be no empirical objects; I interpret this to mean that if there were no cognitive subjects with our forms of intuition, none of the actual empirical objects would exist. In the third passage, Kant claims that empirical spatial objects exist in virtue of being experienced. I take it that this also entails that if ‘we’ were not there to experience them—and by ‘we’ I mean beings with discursive intellects and spatiotemporal forms of intuition—none of the actual empirical spatial objects would exist.20

Collectively, these texts raise the following modal indiscernibility argument against the One Object reading:

  1. It is possible that: t exists and there are no cognitive subjects with a spatial form of intuition.
  2. It is not possible that: a exists and there are no cognitive subjects with a spatial form of intuition.21
  3. at

where ‘a’ is an empirical name, and ‘t’ is a non-empirical name.

The texts cited earlier can be read differently, so that they do not support the modal indiscernibility argument. On this alternative, deflationary reading Kant is not claiming that appearances would not exist if there were no cognitive subjects. He is claiming that these objects would not be appearances if were no cognitive subjects. If there were no subjects to appear to, these objects, which are numerically identical to things in themselves, would not appear. Since the One Object reader understands the distinction between the appearance and the thing in itself as a distinction in two ways of considering the object, it is potentially open to the One Object reader to understand these passages as making the anodyne claim that: if there were no cognitive subjects, a given object could not be considered as it appears because it would not appear to anyone. This alternate reading is strengthened by Kant's claim in the first passage that “as appearances they cannot exist in themselves, but only in us.” On the deflationary reading, this means that appearances cannot ‘exist as appearances,’ i.e. appear, without cognitive subjects to appear to. On my reading, this means that because they are appearances empirical objects cannot exist without subjects to appear to.

However, I think the deflationary reading is not ultimately defensible. In those passages Kant does not make the relatively modest claim that appearances would vanish if there were no cognitive subjects whatsoever; he makes the more striking claim that they would vanish if there were no cognitive subjects with our spatiotemporal form of intuition. In the second passage he writes that the corporeal world “is nothing but the appearance in the sensibility of our subject”; in the third passage he writes that things in space, our a priori form of intuition, exist in virtue of being represented. In the first passage I take it that “the subjective constitution of the senses in general” refers to our specifically spatiotemporal form of intuition; Kant is claiming that these objects would not exist, not only if there were no cognitive subjects whatsoever, but that they would not exist if there were no beings with our form of intuition. Kant is making a modal claim about what would be the case if there were no cognitive subjects with spatial intuition.

The deflationary reading has to understand ‘existence’ in these passages as meaning ‘existing as an appearance’ or ‘exists and appears.’ If ‘exists’ in these passages means ‘exists and appears’ then premise (2) in the modal indiscernibility argument should be understood as:

  • (2*)
    It is not possible that: a exists and appears and there are no cognitive subjects with a spatiotemporal form of intuition.

But if a is numerically identical to a thing in itself this modified premise violates Kantian strictures on our knowledge.22 Even if there were no spatially intuiting cognitive subjects, it does not follow that a would not appear; it might appear to other, non-human cognitive subjects with a non-spatiotemporal form of intuition.23 The deflationary reading avoids the modal indiscernibility argument, but only at the cost of violating Kantian restrictions on our knowledge.

One Object readers may feel, nonetheless, that a deflationary reading of these passages is possible. I have two further points to make about this. First, the point of the second passage from above is to contrast Kant's view with a materialist view on which the existence of the thinking subject depends upon the existence of the bodies of which that subject is constituted. On the materialist view, if there were no bodies, there would be no thinking subjects. Kant's point is that, on his view, the converse dependence relation holds if there were no appropriately constituted thinking subjects, there would be no bodies. This militates in favor of the straight-forward modal reading I gave earlier, rather than a sophisticated deflationary reading. Secondly, I will show in section three that the One Object view is in fact compatible with my reading of these passages; the One Object reader can defeat the modal indiscernibility argument I just gave. I suggest, therefore, that the best strategy for the One Object reader to pursue is to accept that these passage mean what they appear to mean: if there were no human cognitive subjects, none of the actual appearances would exist.

My reading of these passages might also be challenged by offering a de dicto rather than a de re reading. On my interpretation, Kant is making a de re claim about each appearance: if there were no spatiotemporally intuiting cognitive subjects, this very object would not exist. Formally, this is represented by the fact that a singular term, a name for the appearance, appears within the scope of the modal operator in (2). On the de dicto reading, (2) would be replaced by:

  • (2*)
    It is not possible that: there are no cognitive subjects with a spatiotemporal form of intuition and there are appearances.

But this is not something we are in a position to know, for the reasons just mentioned. If anything, Kant would have in mind,

  • (2**)
    It is not possible that: there are no cognitive subjects with a spatiotemporal form of intuition and there are spatial appearances.

If this is what Kant is claiming in these passages, then the modal indicernibility argument is invalid. But Kant seems to be claiming something stronger in these passages. He says the whole corporeal world “would disappear” if there were no cognitive subjects with a spatiotemporal form of intuition; if (2**) were correct, then he would entitled only to claim that the entire corporeal world would not be spatial, hence would not be corporeal. Similarly, in third passage he says that things in space “exist” in virtue of being represented in space; he does not claim merely that they are spatial or exist ‘as spatial’ in virtue of being represented in space, as he should if (2**) were the correct reading. Thus, although (2**) is a coherent response to the modal indiscernibility argument, I think it does not adequately fit the text. As I mentioned above, I do not think this should overly trouble the One Object reader. I show below how the One Object reader can adopt what I have argued is the more straight-forward reading of these passages and rebut the Modal Indiscernbility argument.24, 25

The One Object response to the simple indiscernibility argument (either with definite descriptions or with names) does not suffice as a response to the modal indiscernibility argument. This is because the ‘property’ which (2) ascribes to the appearance is not one that lends itself to being described as a property that the object ‘appears’ to have. The reason for this is that (1) and (2) are really not claims about predicates the object might satisfy, but the identity relation. They could be reformulated as:

  1. ◊(p & ∃x(x = t)
  2. ¬◊(p & ∃x(x = a)
  3. at

where ‘p’ stands for the judgment that there is no cognitive subject with a spatiotemporal form of intuition. The One Object response to the simple indiscernibility argument will not work here because it is not clear what it means that the object appears to be such as to not possibly exist if human cognitive subjects lacked a spatial form of intuition. This does not seem like the kind of a property an object can appear to have.26

The One Object solution to the simple indiscernibility argument sketched earlier cannot solve the modalized indiscernibility argument because that argument is ultimately a problem about the identity conditions of appearances and things in themselves: it shows that appearances and things in themselves have different modal or ‘counterfactual’ identity conditions. This point is made even more explicit by what I'm going to call the ‘individuation’ argument, recently put forward by Ralph Walker.27 The individuation argument is different from the other indiscernibility arguments in that it relies on epistemic premises, premises about our knowledge of the identities of objects, and derives an unacceptable epistemic conclusion about our knowledge of the identities of things in themselves. However, in section three I will show that it admits of a similar solution to the other indiscernibility arguments, thus justifying its inclusion with them.

(c) The diachronic indiscernibility argument

We can know through experiences that distinct appearances are distinct. According to the One Object view, appearances are numerically identical to things in themselves. It follows that our knowledge of the identity and difference of appearances entails knowledge of the identity and difference of things in themselves. For instance, let ‘a1’ and ‘a2’ be names for one and the same appearance, observed at different times. For instance, in Kant's famous example of the ship going downstream they might refer to the ship when it is upstream and when it is downstream, respectively. Let ‘t1’ and ‘t2’ be non-empirical names for the things in themselves that are identical to a1 and a2. Let ‘K’ stand for the sentence operator ‘it is known by us that’. Then consider the following argument:

  1. K (a1 = a2)
  2. a1 = t1 & a2 = t2
  3. ∴ K(t1 = t2)28

Of course the conclusion (3) does not follow unless we strengthen (2) to:

  • (2*)
    K(a1 = t1 & a2 = t2)

But, since (2) is simply the statement of the One Object view, (2*) is the claim that we know that this view is correct,. The One Object reader could deny this premise. But remember that the One Object reading is a claim about what Kant meant by ‘appearance’ and ‘thing in itself.’ It is very implausible that whatever Kant meant by these terms, he was only rationally warranted in believing that they mean what he thought, but did not know what they mean; these are technical terms introduced by Kant himself. Surely if the One Object view is correct, then, on Kant's view, we know what these terms mean. And if we know what these terms mean, then we know the original premise (2) is true.29

The reason this is an argument against the One Object view is that the conclusion, (3), appears to violate Kant's denial that we can have knowledge of things in themselves. What the argument brings out is that, on the One Object view, our empirical knowledge of the identities of empirical objects translates into knowledge of the identities of things in themselves. After all, if (3) is true, we know the following to be true:

  • (4)
    The thing in itself that appears as the ship at time 1 is identical to the thing in itself that appears as the ship at time 2.

because the non-empirical names t1 and t2 were introduced to refer to the thing in itself that appears as the ship at time 1 and to the thing in itself that appears as the ship at time 2, respectively.

The One Object reader might simply dig in his heels and claim that Kant's strictures on knowledge of things in themselves allow us, nonetheless, to know that (4) is true; after all, he might claim, knowing how to individuate empirical objects still leaves us devoid of substantive knowledge of their intrinsic natures or how they are ‘in themselves.’ I do not think this is a satisfactory response; knowledge that the thing in itself appearing as this ship now is the same thing in itself that appears as this ship later when it is further downstream seems like exactly the kind of substantive knowledge claim about things in themselves Kant meant to deny. However, as I show in section three, there is a better alternative available to the One Object reader, one that does not bring with it the consequence that we have such knowledge. Consequently, I am not going to argue further that this kind of knowledge violates Kant's doctrine that we are ignorant of things in themselves.

Finally, the One Object reader might be tempted to claim that we do not know premise (2) in the original argument because (2*) is meaningless. Identity statements involving an empirical name and a non-empirical name, it might be claimed, lack any sense whatsoever because there is no means, empirical or non-empirical of verifying them. But this response simply will not do, because the One Object view centrally consists in making precisely such an identity claim: the thing in itself is identical to the appearance. The One Object reader makes this response at the cost of rendering his own view unsayable.

2. Monism and Indiscernibility

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

Before proposing how the One Object reading can surmount these challenges, I want to draw some parallels between that reading and the problems it faces, and a structurally similar view in material constitution, Monism, and the problems it faces. In the next section I'll show that the One Object reading can learn from the Monist how to respond to the indiscernibility and individuation arguments from the previous section.

According to the One Object view, the appearance and the thing in itself are one and the same object, but they have different spatial and modal properties, and (for all we know) different cross-temporal persistence conditions.30 Seen in this way, the One Object view is similar to what, in this paper, I will call ‘Monism,’31 the view that ordinary material objects are identical to the matter of which they are made.32 On the Monist view, the sword is identical to the steel that makes it up, just as a statue is identical to the clay of which it is made. However, the Monist faces challenges arising from the fact that ordinary material objects seem to differ in certain of their properties from the matter of which they are made.

Consider an alternate version of Allan Gibbard's famous case of Lumpl and Goliath.33 In the morning I pour liquid copper and tin into the mold of a statue of Mickey Mouse. As the molten metal cools, it forms a quantity of bronze and a statue. Call the bronze ‘Lumpl’ and the statue ‘Goliath.’ Later that evening, I drop Lumpl/Goliath into a vat of acid that instantly dissolves Lumpl into its constituents, thus simultaneously destroying Goliath and Lumpl. Lumpl and Goliath are spatially co-located for the entire duration of their existence, so, intuitively, there is some force to the claim that they are one and the same object. Consider the following argument that they are distinct objects:

  1. If I were to smash Goliath at noon, Lumpl would not cease to exist.
  2. If I were to smash Goliath at noon, Goliath would cease to exist.
  •     (LL)
    (x)(y)(x = y ⊃ Fx [LEFT RIGHT ARROW] Fy), for any F
  •     3.
    Lumpl ≠ Goliath.

Monists have developed various strategies for resisting such modal indiscernibility arguments. Getting into the details of various Monist proposals is not appropriate in this context, because ultimately we want to learn from Monism how to defend the One Object reading of Kant. Instead, I am going to describe one very general Monist solution, modeled on the work of Michael Jubien, and then apply that strategy to the One Object reading.

What this argument brings out is that, if the Monist wants to maintain that (1) and (2) are true, then she must deny that de re modal contexts like (1) and (2) are referentially transparent34; otherwise, by substitution of identicals, (1) and (2) must have the same truth-value. (1) and (2) do not attribute and deny, respectively, one and the same property to an object; which property they attribute, or deny, to that object depends upon how that object is picked out or named. The challenge to the Monist is to explain why (1) and (2) are referentially opaque contexts. Call some possible state-of-affairs a ‘witness’ to (1) just in case both the antecedent and consequent of (1) would have been true if that state-of-affairs were actual; likewise for (2). Let o be the object that is actually Goliath. What the Monist needs to deny is that the witnesses for (2) are states-of-affairs in which o ceases to exist. In other words, the Monist needs to deny that the witnesses for (2) are states-of-affairs in which the object that is actually Goliath ceases to exist. There are states-of-affairs, the Monist must concede, in which o continues to exist but in which Goliath does not. In other words, for some state-of-affairs to be a state-of-affairs in which Goliath exists is not a matter of the object that is actually Goliath existing. But this is incoherent if we take the property of being Goliath to be the property of being numerically identical to o where o is the object that is actually Goliath. Hence, the Monist must deny that these are the same property, if she wants to adopt the proposed strategy. Now I will explain how the Monist can coherently deny this.

On the Monist view, every material object (i.e. every object constituted solely by matter) is numerically identical to some quantity of matter. Some of these objects instantiate ordinary sortal kinds, like statue, suit of armor, or cellular tower; in virtue of this, these quantities of matter are statues, suits of armor and cellular towers. However, whether, in some counterfactual state-of-affairs, this very suit of armor or this very same cellular tower exists is not a matter of whether, in that state-of-affairs, there is a quantity of matter numerically identical to the quantity of matter that is actually this suit of armor or this cellular tower. As the Lumpl and Goliath example brings out, the quantity of matter that is actually some ordinary empirical object can exist, even though that ordinary empirical object ceases to exists. After being smashed, Goliath would cease to exist but the quantity of matter that is actually Goliath would continue to exist. This means that for any name of an ordinary empirical object falling under a standard sortal kind K, we can distinguish between the object this name refers to, which is some quantity of matter, and the property of being that K. For instance, if ‘Goliath’ is introduced as the name of a statue, its referent is some quantity of matter, o. But the property of being Goliath, or, equivalently, being that very statue, is not the property of being numerically identical to o. As Gibbard's argument brings out, the quantity of matter o that is actually Goliath—that has the property is Goliath—might exist in some counterfactual circumstance in which no object has the property is Goliath.35

Thus the Monist can resist the modalized indiscernibility argument by reinterpreting its premises as:

  • (1*)
    If I were to smash Goliath at noon, there would still be an object that is Lumpl.
  • (2*)
    If I were to smash Goliath at noon, there would be no object that is Goliath.
  • (3)  
    Lumpl ≠ Goliath

The conclusion clearly does not follow, because the property being Goliath is not the same as the property being Lumpl, i.e. the property of being numerically identical to the object o that is actually Goliath.

This Monist solution to the modalized indiscernibility argument can be generalized into a solution to a cross-temporal indiscernibility argument. After creating Lumpl and Goliath at time t some time passes, and a piece of Lumpl is annihilated. I replace the piece with some new bronze. Call the statue that results at time t* ‘New Goliath’ and the matter that results ‘New Lumpl.’ Consider the following argument:

  1. Lumpl ≠ New Lumpl
  2. Goliath = New Goliath
  3. New Goliath = New Lumpl
  4. ∴Lumpl ≠ Goliath

Premises (1) and (2) are true because, intuitively, statues can survive changes in non-essential material parts while quantities of matter cannot. The Monist is committed to premise (3) because this is just the thesis of Monism applied at a later time in the life of Goliath.

The Monist can easily resist this conclusion of this argument by reinterpreting its premises as follows:

  1. Lumpl is numerically distinct from New Lumpl.
  2. The object that is Goliath at t is the same statue as, but not numerically identical to, the object that is New Goliath at t*.
  3. The object that is New Goliath at t* is identical to New Lumpl.

From which, clearly, it does not follow that

  •     4.
    The object that is Goliath at t is numerically distinct from Lumpl.

The Monist can resist the conclusion of the cross-temporal indiscernibility argument because the Monist distinguishes between the property being the same statue as Goliath from the property being identical to the object that is presently Goliath, that is, the property being Lumpl.

This Monist solution consists, ultimately, in distinguishing numerical identity from the contingent, sortal-relative, identity relation being the same K as where K is an empirical sortal term. It does not commit the Monist to denying that numerical identity is necessary and sortal-independent; it only commits the Monist to the claim that the relation being the same K can hold between numerically distinct objects and do so contingently, and to the claim that many apparent claims of numerical identity are disguised claims of contingent identity.36 In the next section, I extend this strategy to the One Object reading of Kant's transcendental idealism.

3. Monism and the One Object Reading

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

(a) Structural parallels with Monism

The Monist holds that an ordinary material object is numerically identical to the matter of which it is made. According to the One Object view, an appearance, a spatiotemporal empirical object, is numerically identical to the non-spatiotemporal thing in itself that appears as that appearance. The Monist faces the problem that, intuitively, ordinary material objects have different modal and temporal persistence conditions than the matter of which they are made. This generates a modal and a temporal indiscernibility argument against Monism. Similarly, as we saw in section one, Kant ascribes different modal properties to the appearance and to the thing in itself: the thing in itself would exist even if human cognitive subjects had a non-spatiotemporal form of intuition, while the appearance would not. The parallels between Monism and the One Object reading, and between the indiscernibility arguments against them, suggest that the proposed Monist solution to these arguments generate parallel One Object solutions to the corresponding arguments. I explore and develop those parallel One Object solutions in this section.

The Monist can solve the modal and cross-temporal indiscernibility arguments, and thereby accommodate the intuitions on which they are based, by distinguishing between, for instance, the property of being numerically identical to the object that is actually or presently Goliath and the property of being Goliath or, equivalently, the property of being the same statue as Goliath. It is possible for the object that is actually Goliath to exist, even though no object is Goliath, or, equivalently, no object is the same statue as Goliath. Similarly, some later object may be Goliath, even though the object that is presently Goliath ceases to exist; the later object has the property of being the same statute as the original object, but it is not numerically identical to it.

(b) How to answer the modal indiscernibility argument

The One Object reading can adopt a similar strategy. Recall the modal indiscernibility argument from section one:

  1. It is possible that: t exists and there are no human cognitive subjects with a spatial form of intuition.
  2. It is not possible that: a exists and there are no human cognitive subjects with a spatial form of intuition.
  3. at

Let us reformulate the premises to bring out the centrality of identity to this argument. The first premise should be reformulated as:

  • (1*)
    It is possible that: there is an object that is numerically identical to t and there are no human cognitive subjects with a spatial form of intuition.

If we reformulate the second premise as:

  • (2*)
    It is not possible that: there is an object that is numerically identical to a and there are no cognitive subjects with a spatial form of intuition.

Then the argument is valid and entails the One Object view is false. However, if we reformulate the original second premise as:

  • (2†)
    It is not possible that: there is an object that has the property of being a and there are no cognitive subjects with a spatial form of intuition.

the One Object reader can resist its conclusion. The One Object can resist the conclusion, however, only if she denies that the property of being a is the same as the property of being identical to the object that is actually a, because then (2†) reduces to (2*) and the argument validly entails that ta.

The key to the One Object solution to the Modal indiscernibility argument is to understand the property of being a as the property of being the same appearance as a rather than the property of being numerically identical to the object that is actually a. But this is very natural on the One Object picture. It means the One Object reader has to think that being a particular appearance, for instance, being this suit of armor, is not a matter of being numerically identical to the thing in itself that appears as this suit of armor. Something distinct from that thing in itself could appear as this suit of armor and thus be that suit of armor in virtue of being the same appearance as that suit of armor. Being this suit of armor is a matter of bearing the spatiotemporal and causal relations to this suit of armor in virtue of which we individuate appearances across time and through counterfactual situations. A fully worked out One Object reading of Kant would have to describe precisely what these individuating conditions are, but that lies outside the scope of this paper.

(c) How to answer the diachronic individuation argument

The same strategy can be applied to solve the cross-temporal individuation argument against the One Object view. Recall from section one:

  1. K (a1 = a2)
  2. K (a1 = t1 & a2 = t2)
  3. K (t1 = t2)

where ‘a1’ and ‘a2’ are names for one and the same ship when it is upstream and when it is downstream, respectively, ‘t1’ and ‘t2’ are non-empirical names for the things in themselves that appear as a1 and a2, respectively, and ‘K’ stands for the sentence operator ‘it is known by us that’.37 The One Object reader can reinterpret premise (1) as follows:

  • (1*)
    K(a1 is the same appearance as a2)

which renders the conclusion invalid. Intuitively, the boat seen downstream may be the same appearance as the boat seen upstream, but the thing in itself appearing as the boat seen downstream is not numerically identical to the thing in itself appearing as the boat seen upstream. Consequently, our knowledge that the former boat is the same appearance as the later boat does not entail that we know that they are appearances of one and the same thing in itself.38

The common theme in the One Object and Monist solution to these indiscernibility arguments is to avoid what Michael Jubien calls ‘object fixation.’39 We are inclined to think that in settling issues about the de re modal properties of a particular object we use a name to refer to that object and then consult counterfactual situations or ‘possible worlds’ in which that very object exists to determine which properties it has accidentally and which necessarily. In the case of Monism, this is equivalent to thinking that ordinary empirical objects only have the de re modal properties of the matter of which they are composed. In the case of the One Object view, this is equivalent to thinking that empirical objects have exactly the modal properties of the things in themselves to which they are identical. The solution in both cases consists in freeing ourselves from ‘object fixation.’ In the Monist case this requires recognizing that the de re modal properties of an ordinary material object of kind K may be determined by the properties of objects that are that K in different counterfactual situations or ‘possible worlds.’ In the One Object case, it requires recognizing that the de re modal properties of appearances are determined by the properties of objects that are the same appearance as that object in different counterfactual situations or ‘possible worlds.’40

The One Object solution to these problems that I have sketched consists in divorcing the identity and difference of appearances from the numerical identity and distinctness of the things in themselves to which they are numerically identical. It does this by refusing to analyze what you might call ‘empirical object-essences’ (following Jubien), that is to say, properties like being this very appearance, as involving numerical identity. Instead, it takes the property of being the same appearance as a property whose application conditions are given by our spatiotemporal and causal criteria of individuation. Every appearance is numerically identical to a thing in itself, but being that appearance is a matter of bearing certain spatiotemporal and causal relations to that object, not of being numerically identical to that thing in itself. One clear consequence of this is that judgments that appear to be judgments of numerical identical, e.g. ‘this ship is the same ship that I saw upstream a moment ago’ are not really judgments of numerical identity at all.

This is precisely parallel to the Monist strategy of divorcing identity for objects of ordinary empirical kinds (i.e. kinds other than quantity of matter) from numerical identity. On the Monist view, when we ask whether the statue that results from replacement of a small part of Goliath is the same statue we are not asking whether the object that results is numerically identical to the original object (it is not); we are asking whether it bears the contingent, sortal-relative relation being the same statue as to it. Just as the Monist need not deny that numerical identity is necessary and sortal-independent, the One Object reader need not deny this either. The One Object reader need only claim that the relation that we typically express when talking and thinking about ‘the same’ empirical object is not numerical identity, but the contingent, sortal-relative relation is the same appearance as.

4 Why the One Object Reading Nonetheless Fails

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References

(a) The synchronic individuation argument

In section one I raised a cross-temporal identity argument against the One Object reading. In a recent paper, Ralph Walker raised the same argument against the One Object view.41 Walker intended to raise a synchronic identity problem as well. I shall now set out the synchronic problem and show that it is more damaging to the One-Object view than the diachronic problem.

There is presently a suit of armor—call it ‘a1’—and a sword—call it ‘a2’—in my parlor. Let ‘t1’ refer to the thing in itself that appears as the suit of armor now and ‘t2’ refer to the thing in itself that appears as the sword now.42 Consider the following argument:

  1. a1 is not the same appearance as a2. [Premise]
  2. a1 = t1 & a2 = t2. [Premise]
  3. a1 is the same appearance as a1. [Premise]
  4. t1 appears as the same appearance as a1. [From (3) by substitution.]

Note: the form of the judgment has changed from (3) to (4), when ‘t1’ is substitued for ‘a1.’ This is just the idea, familiar from section one, that the form of the judgment must change when a non-empirical name is substituted for an empirical one, e.g. ‘a is F’ becomes ‘t appears F’ when the non-empirical name ‘t’ is substitued for ‘a.’

  •     5.
    t2 does not appear as the same appearance as a1. [From (1) by substitution; see previous note]
  •     6.
    t1≠t2. [From (5) and (6)]

All of the premises of this argument are things which, according to the One Object view, we know to obtain: we know that the suit of armor and the sword are different appearances, we know the One Object reading, (2), is true, we know that that (3) is true by the definition of the relation is the same appearance as. If this is correct, then we know the conclusion, (6), is true. In other words, on the basis of knowing that two appearances are distinct, we know that the things in themselves that appear as those objects now are distinct. It appears that, contrary to my best attempts to save the One Object view, that view is still committed to attributing to us a knowledge of the identities of things in themselves that we cannot possess.

We should be clear about why this problem does not admit of the same solution as the problem of diachronic identity. The solution to the problem of diachronic identity was to show that the following might be the case: the thing in itself that is (appears as) the ship at time T is distinct from the thing in itself that is (appears as) the ship at time T*, despite the fact that those things in themselves are (appear as) the same appearance. That path is blocked in this case, because there is only one time in question, the present moment. Translated into the synchronic case, the move that saved the One Object view in the diachronic case would be: the thing in itself that appears as the suit of armor and not the sword at T is identical to the thing in itself that appears as the sword and not the suit of armor at T*( = T). In the synchronic case, the One Object view requires attributing to one and the same thing in itself contradictory properties: being and not being the same appearance as the suit of armor.

It should be noted that this argument only entails individuative knowledge of things in themselves in one direction: if there are two distinct appearances at one time, they must be appearances of distinct things in themselves. It does not entail that there cannot be two distinct things in themselves that appear as one and the same appearance at a given time. Thus, it does not entail that there is a one-to-one correspondence between appearances and things in themselves that ‘appear’; it does not, therefore, entail that we can count things in themselves by counting appearances. However, it does entail that we have substantial knowledge of the identities of things in themselves: any two distinct appearances at a given time are appearances of distinct things in themselves. And this entails that we can know that there are at least as many things in themselves as there are appearances, as I will now argue.

I want to convince you that this is a serious problem for the One Object view, so serious that the One Object view should be given up rather than admit that we can have such knowledge of things in themselves. Bodies, for Kant, are infinitely divisible, although not infinitely divided.43 I take this to mean that if a body wholly occupies region R and R* is a sub-region R then that body potentially has a part that wholly occupies R* although it need not actually have such a part. This generates the following argument:

  1. □ (a)(b)(a and b are different appearances ⊃ the thing in itself that appears as a is distinct from the thing in itself that appears as b).
  • Reason: This is the necessitation of the conclusion of the synchronic indiscernibility argument from earlier. I am exploring the consequences, for the One Object view, of simply accepting its conclusion. Furthermore, I am assuming that since the One Object view is a view about what ‘appearance’ and ‘thing in itself’ mean for Kant, it is not merely a contingent fact that appearances and things in themselves are numerically identical. Note, though, that, the One Object view is committed not merely to the truth of (1) but to its knowability.
  •     2.
    Let B be a body that wholly occupies region R.
  •     3.
    Let R1 and R2 be distinct sub-regions of R.
  •     4.
    ◊ (B has a proper part B1 wholly occupying R1 & B has a proper part B2 wholly occupying R2).
  • Reason: I take it this is what Kant means by saying that matter is ‘infinitely divisible’ though not ‘infinitely divided’: it is not necessary that B actually has a part occupying every sub-region of R, but merely that it could have those parts. Now, it might be objected that the infinite divisibility of matter does not mean that B could have parts in any region, no matter how gerrymandered; perhaps it entails only that in every spherical sub-region of R, B could have a part, etc. This concession will not materially affect the argument, I think. No matter what qualification we add on the sub-regions R1 and R2, if we add the same qualifications to the rest of the premises and the conclusion the argument remains valid. Even that qualified conclusion is, I think, unacceptable.
  •     5.
    □ (B1 and B2 occupy distinct sub-regions of R ⊃ B1 is a distinct appearance from B2).
  •     6.
    □ (B1 is a distinct appearance from B2 ⊃ the thing in itself that appears as B1 is distinct from the thing in itself that appears as B1).

Reason: this follows from (1).

  •     7.
    ◊(B has a part B1 wholly occupying region R1 & B has a proper part B2 wholly occupying region R2 & the things in themselves that appears as B1 and B2 are numerically distinct).
  • Reason: B, B1 and B2 are distinct appearances by (5) because they occupy distinct sub-regions of R. By (6) the things in themselves that appear as these objects must all be distinct.
  •     8.
    If x≠y then □ x≠y.
  •     9.
    Whether B is divided or not, the things in themselves that would appear as its parts if it were divided exist.
  • Reason: which things in themselves exist does not depend upon how or whether we have divided the appearances.

  •    10.
    By (8) and (9), the things in themselves that would appear as the parts of B in R1 and R2, if it were divided, are distinct.
  •    11.
    For every pair of distinct sub-regions of R there are at least two things in themselves.

But this means that there are at least as many things in themselves as there are regions of space. While these things in themselves need not have mereological structure that mirrors the mereological relations of regions of space—thus, strictly speaking, this does not show that things in themselves are divided the way bodies are—it does show that there is at least an uncountable infinity of things in themselves.44 Furthermore, it shows that we can know that there is an uncountable infinity of things in themselves, for we can know all of the premises of this argument. This strikes me as obviously in violation of Kant's restriction on knowledge of things in themselves. To the One Object reader who stubbornly insists that this is compatible with ‘noumenal ignorance’ I have no further arguments to give, only an incredulous stare. I am going to continue on the assumption that this argument shows that the One Object reader cannot simply accept the conclusion of the synchronic indiscernibility argument, but must find some way of either rejecting one of its premises or showing it to be invalid.

(b) How to answer the synchronic individuation argument?

The challenge to the One Object reader is how to understand the following state-of-affairs: a single thing in itself t appears as two distinct appearances a and b at the same time, that is, has both the property is the same appearance as a and is not the same appearance as a. One move the One Object reader might make is to introduce some complexity into the thing in itself: t has parts, one of which has the property is the same appearance as a and the other of which has the property is a different appearance from a. However, this violates the One Object view: appearance a would be identical to one part of t while appearance b would be identical to a distinct part of t.

The other way to save the One Object view is to introduce complexity, not into the things in themselves, but into their appearance properties. Here's one way of looking at the problem. The two appearances, the suit of armor and the sword (a1 and a2 respectively) have contradictory properties. For instance, a1 is in location L (let us say) while a2 is not in location L. This leads to a short argument for the distinctness of t1 and t2:

  1. a1 is in location L
  2. t1 appears as located at L
  3. a2 is not in location L
  4. t2 appears as not located in L
  5. ¬ (t2 appears as located in L)
  6. t1t2

Showing how the One Object reader can resist this argument will help us see how he can resist the synchronic individuation argument from above. The One Object reader can resist this argument by denying that (4) entails (5), i.e. by denying that from the fact that some thing in itself appears to lack a property it follows that it does not also appear to have that property. The One Object reader can hold that things in themselves can appear as having a variety of contradictory properties because a given thing in itself can appear as multiple appearances. In the example, one and the same thing in itself can appear as a suit of armor and a sword, and thus appear both to be a suit of armor and not be a suit of armor, because it appears as two simultaneous appearances. The One Object reader simply needs to deny the inference from the claim that the thing in itself appears as not being a suit of armor to the claim that it does not appear as a suit of armor.

In describing these as ‘two appearances’ it may seem that I have already contradicted the very idea of the One Object reading: the appearance is numerically identical to the thing in itself, so if the appearances are numerically distinct, so are the things in themselves. But the One Object reader can hold that one and the same thing in itself t can appear as a1 and a2 where a1 and a2 are distinct appearances (i.e. they do not bear the relation being the same appearance as to one another) although ‘a1’ and ‘a2’ both refer to t. On this modified version of the One Object view, we can say truly both that a1 = t and a2 = t and that a1 and a2 are distinct appearances.

(c) Why this is not in fact a solution

The Modified One Object view, in order to avoid the synchronic identity objection, admits that one thing in itself can have ‘contrary’ appearance properties (e.g. appearing as not F and appearing as F) but no contradictory properties (e.g. appearing as F and not appearing as F). It does so by attributing the contrary appearance properties to different appearances. The Modified One Object view has to distinguish between the (impossible case) in which a thing in itself appears as one appearance that simultaneously possesses contrary appearance properties and the case in which one thing in itself appears as two appearances with contrary appearance properties. In other words, when

  1. The thing in itself t appears to be in location L now and appears to not be in location L now.

the modified One Object view needs to distinguish between the impossible state-of-affairs in which

  1. The thing-in-itself t appears as a, where a is in location L now and not in location L now.

and the possible state-of-affairs in which

  •     b.
    The thing-in-itself appears as a where a is in location L now and appears as b where b is not in location L now.

More generally, the modified One Object view needs to endorse a principle of the following form:

  • (OO)
    For any x and for any F, if x is a thing in itself that appears F-at-t and appears not-F-at-t then there are distinct appearances a and b such that x appears F-at-t qua a and x appears not-F-at-t qua b.

This allows the One Object reader to resist the conclusion of the original synchronic individuation argument. For on this modified One Object view, premise (4) and premise (5) of that argument should be understood as:

  • (4*)
    t1qua a1 appears as the same appearance45 as a1.
  • (5*)
    t2qua a2 does not appear as the same appearance as a1.

which, clearly, do not entail that t1 ≠ t2.

But now the One Object reader is hard pressed to explain the difference between her view and the Two Object view, on which

  • (TO)
    For any x and for any F, if x is a thing in itself that appears F-at-t and appears not-F-at-t then there are distinct appearances a and b such that x appears as a and a is F-at-t and x appears as b and b is not-F-at-t.

The question is what does it mean for a thing in itself to appear F qua a but appear not-F qua b where a and b are distinct appearances? On the Two Object view we have one thing in itself appearing as two numerically distinct appearances, which are themselves numerically distinct from the thing in itself. On the Two Object view, it seems that we have one thing in itself standing in two different appearance relations to two different bundles of properties.

Nonetheless, the One Object reader will insist, the bearers of those properties, the appearances, are numerically identical to the original thing in itself. After all, on the One Object view it comes out true that

  • (I)
    a = t and b = t

But I think this difference is merely apparent. After all, the One Object view makes (I) true by a stipulation on the referent of the empirical names ‘a’ and ‘b.’ It isn't clear that there is any substantive sense in which these appearances are numerically identical to the thing in itself; it is not clear that there is any substantive sense in which this is still a One Object view. After all, if we are going to count appearances, we will count them by the relation ‘being the same appearance as’ rather than numerical identity. Thus, the number of appearances will not be the same as the number of things in themselves that appear. Secondly, our individuation of the appearances does not track identity of things in themselves; that is clear from the One Object solution to the diachronic and synchronic identity problems. If anything, on the One Object view appearances seem to be numerically identical to clusters of properties the thing in itself appears to have. Empirical names are indices for the appearance relation that holds between the thing in itself and these clusters of properties; there is one cluster of properties for every empirical name. This strongly suggests that empirical names either refer to these clusters of properties or refer to objects that possess these clusters of properties, which objects are numerically distinct from things in themselves. But that would entail that empirical names for different appearances (e.g. names ‘a’ and ‘b’ such that it is true that a and b are different appearances) actually refer to numerically distinct objects, either bundles or bearers of properties. The One Object view remains distinct from the Two Object view only by making a questionable terminological stipulation. In its substance, the modified One Object view does not differ from the Two Object view.

Some readers might object that, rather than showing the superiority of the Two Object view, this merely deflates the difference between that view and the One Object view. After all, haven't I just argued that a properly modified One Object view is a mere ‘terminological variant’ of the Two Object view? Yes, but our understanding of this modified One Object view is parasitic on our understanding of the Two Object view. The Two Object view has not been modified in any way to make it more like the One Object view. The One Object view has transformed itself, under the pressure of rendering it internally coherent and compatible with Kant's doctrine noumenal ignorance, into a view that differs from an unaltered Two Object view only by a questionable terminological stipulation. This ‘deflates’ the difference between the One and Two Object views only in the sense that it shows that the former view can distinguish itself from the latter view at the price of becoming untenable. Hence, I take this paper to have argued in favor of the Two Object view.

I began this paper by raising several objections against the Metaphysical One Object view. I showed how the Metaphysical One Object view could resist these objections by adopting strategies developed by Monists about material constitution in order to resist similar objections. This resulted in a more sophisticated form of the Metaphysical One Object reading. However, in this section, I argued that these strategies cannot be adapted to resist the synchronic identity problem: our knowledge of the identity and distinctness of appearance at a given time seems to entail that we have a kind of knowledge of things in themselves that we cannot have. The only way for the Metaphysical One Object view to respond to the synchronic identity problem is to endorse a ‘modified’ version on which one and the same thing in itself can appear as multiple appearances. But this modified version, I have argued, is only verbally different from the Two Object view. In other words, in response to philosophical pressures, the Metaphysical One Object view itself transforms into the Two Object view; the best, most defensible version of the One Object view is indistinguishable from the Two Object view. This, I take it, is strong confirmation that the Two Object view is superior. However, it is not decisive confirmation. If the textual evidence decisively weighs in favor of the One Object view, then we must attribute to Kant a view that, as I have shown, is incompatible with the restriction of theoretical knowledge to empirical objects. However, this is quite unlikely. Given the variety of texts that support a Two Object interpretation and the lack of a One Object view that respects very high-level features of Kant's theoretical philosophy without being merely a terminological variant of the Two Object view, it is very likely that, in the end, the balance of reasons, both textual and philosophical, will favor the Two Object view.

Notes
  1. 1

    I would like to thank Tobias Rosefeldt, Karl Schafer, Rolf Horstmann, Elijah Chudnoff, Amie Thomasson, Lucy Allais, and Ralf Bader for helpful comments on this paper. I presented earlier versions of it at the Colloquium for Classical German Philosophy at Humboldt Universität zu Berlin, and to the Faculty Colloquium at the University of Miami; thanks to both audiences for informative discussions. I am especially grateful to an anonymous referee for this journal, who provided very detailed and genuinely insightful comments on earlier versions.

  2. 2

    E.g. the interpretation offered by Langton (1998) does not fit easily into either camp because her official view is that appearances are not ‘objects’ at all but relational properties of things in themselves, substances with intrinsic properties. This could be developed either in a one object direction (talk of appearances is just talk of substances qua their relational properties) or in a two object direction (appearances are properties of substances, and thus are numerically distinct from the substances in which they inhere). The now standard distinction between One Object readings and Two Object readings is borrowed from Ameriks (1982).

  3. 3

    Although, strictly speaking, we do not know on theoretical ground whether there is a plurality of things in themselves, or just one.

  4. 4

    See section one for textual evidence that, on Kant's view, the existence of appearances does depend upon our experiencing them. However, there is also putative textual evidence against this interpretation; for instance, Kant's claim that “representation in itself does not produce its objects in so far as existence is concerned” (A92/B125). But note that Kant here says that our representations are not a sufficient condition of the existence of their objects; he does not say they are not a necessary condition.

  5. 5

    On this issue, see Adams (1997), Aquila (1979) and Ameriks's discussion of Aquila in his (1982).

  6. 6

    Influential Epistemological interpretations include Bird (1962), Prauss (1974), Allison (1983/2004) and Collins (1999).

  7. 7

    I want to make clear that I am not claiming that, on the Epistemological reading, Kant's argument for the ideality of space is sound. In fact, I think it is unsound. To consider an object as a thing in itself is to abstract from the sensible conditions of human knowledge. I think it follows that if space is a sensible condition of knowledge, then things in themselves are neither determinately spatiotemporal nor determinately not spatiotemporal. Similarly, if I abstract from the gender of a male job candidate I consider the candidate as neither determinately male nor determinately female; I do not consider him as female! But Kant's conclusion is not that things in themselves are neither determinately spatiotemporal nor determinately non-spatiotemporal; his conclusion is that things in themselves are determinately non-spatiotemporal. Consequently, I think that, on the Epistemological reading, Kant's argument is unsound. This criticism of the Epistemological reading is made by Guyer (1997), 279–329; and Van Cleve (1999), 8.

  8. 8

    This would entail that appearances are numerically distinct from things in themselves if we took ‘x qua F’ talk to refer to a special class of entities, so-called ‘qua objects.’ On such view, Socrates-qua-sitting is a distinct object from Socrates; for instance, Socrates-qua-sitting comes into existence when, and only when, Socrates sits, and ceases to exist thereafter. See Fine (1982) for a theory of qua objects. However, when I say that for the One Object reader the appearance is the object qua phenomenal properties, I take this to mean that: judgments about the appearances are judgments about the object qua its phenomenal properties. Thus, ‘qua’ modifies the predication, not the object.

  9. 9

    Cf. Adickes (1924), Rosefeldt (Ms.), Allais (2004), Langton (1997), McDaniels (Ms). Some readers might balk at the inclusion of Adickes among the One Objet readers; see, however, Adickes (1924), 20, 27.

  10. 10

    E.g. Allison (2004), 19, 48.

  11. 11

    For an influential critique of the Epistemological reading, see Van Cleve (1999), 144–6.

  12. 12

    Formally: (x)(y)(x = y ⊃ Fx [LEFT RIGHT ARROW] Fy) for any property F.

  13. 13

    By which I mean that the thing in itself that appears as a given appearance is numerically identical to that appearance. In conversation, I have noted that some One Object readers deny that they are committed to claims of numerical identity. First of all, if the One Object reader is not committed to numerical identity, the name ‘One Object reading’ is extremely misleading (and I didn't invent it!). Secondly, and more seriously, if the metaphysical One Object reader gives up the claim of numerical identity, their view becomes a version of the Two Object view. Thirdly, it is easy to find passages where Metaphysical One Object readers explicitly commit themselves to numerical identity: Adickes (1924), 20, 27; Allais (2004), 657; Langton (1997), 13; McDaniels (ms); Westphal (1968), 120.

  14. 14

    I take up this project in Stang (ms). Some commentators have claimed that Kant changed his mind from the A edition (where he favored a Two Object view) to the B edition (which supposedly expresses a One Object view), or that it is a mistake to assume that at any given period he had a settled view on the matter. I argue against this interpretation in Stang (ms).

  15. 15

    Why can't the One Object reader deny that the definite description ‘the thing in itself that appears as the appearance that is ϕ’ refers? The One Object view cannot consistently maintain (a) her One Object view, (b) the claim that there is a unique appearance that is ϕ, and deny that (c) there is a unique thing in itself that appears as that appearance. To do so deny (c) would be to immediately transform the view into a version of the Two Object view.

  16. 16

    Cf. Langton (1997).

  17. 17

    The One Object reader might also consider a more straight-forward move: simply deny (1). The One Object reader might claim that (1) is true, although not assertible (in normal contexts). The idea would be that any judgment about an object using a ‘non-empirical name’ (a name used to abbreviate a definite description like ‘the thing in itself that appears as the ϕ-appearance’) carries the conversational implicature that the property being ascribed to the object is a property the object has ‘in itself,’ where having a property ‘in itself’ can be glossed as having the property intrinsicially or having it non-response-dependently, etc. While this is interesting, it seems deeply un-Kantian to me. It would entail that the thing in itself that appears as the table is spatial, although asserting this carries the misleading implication that spatiality is a property the object has ‘in itself.’ But Kant does not claim that it is misleading to describe things in themselves as spatial; he repeatedly and adamantly denies that they are spatial. Thanks to an anonymous referee for pressing me on this point.

  18. 18

    I add the qualification ‘reference-fixing’ because I do not intend to suggest that the name is semantically equivalent to the definite description; the definite description is used to fix the referent, but otherwise is not part of the semantic value of the name. This idea derives from Kripke (1980).

  19. 19

    Cf. A490–1/B518–9, A520/B492-A521/B493, A494/B522, Ak. 4:354, R 5086, and R 5109.

  20. 20

    Thus I take ‘our’ in the first passage, and ‘the thinking subject’ in the second passage to refer not just to any thinking subject, nor to specifically human subjects, but to subjects with our most basic intellectual capacities: discursive intellects with spatiotemporal forms of intuition. Kant cannot be claiming that appearances depend upon specifically human beings, for surely the same appearances would exist for non-human martians that nonetheless cognize the world in the same way human beings do. On my reading, these martians would be included in the scope of ‘we.’ Similarly, Kant cannot be claiming that appearances depend upon the existence of some cognizing subject or other, for, as he makes clear in the third passage, appearances exist in virtue of specifically spatial experiences of them. Any being with a spatiotemporal form of intuition has a fortiori a discursive intellect, so appearances depend upon there being discursive cognizers with spatiotemporal intuition. Thanks to an anonymous referee for pressing me to clarify this point.

  21. 21

    Strictly speaking, in these passages Kant makes a counterfactual claim: where a is an appearance, (there are no human subjects) □ [RIGHTWARDS ARROW] (a would not exist). But I take it that this is not merely a counterfactual claim; it is the necessary ‘or strict’ conditional I have stated in the main text. Kant is not just saying that there would be no such appearance; he is saying that necessarily there would be no such appearance.

  22. 22

    Some readers may notice that I am effectively assuming that modal contexts are referentially transparent with respect to empirical names and observe that by giving up this assumption we might be able save the One Object view from the modal indiscernibility argument. In section three I develop a One Object solution to the modal indiscernibility argument that is equivalent to giving up this assumption.

  23. 23

    See A42/B59, B72, A231/B283 and Prolegomena §57 (Ak. 4:351). All citations to the Critique of Pure Reason use the customary format of giving the page in the 1st-edition of 1781 (A), followed by the page in the 2nd-edition of 1787 (B) (e.g. A327/B384). Citations to other works of Kant give the volume and page number in the Academy edition, Kant (1900-) (Ak.). When followed by a four-digit number, ‘R’ refers to Kant's unpublished Reflections in vol. 17 & 18 of the Academy edition. Unless otherwise noted, translations from the Critique are from Kant (1998).

  24. 24

    Thanks to an anonymous referee for pushing me to clarify this argument.

  25. 25

    In Langton (1998), 209–210 Rae Langton tries to explain away the apparently idealist consequneces of Kantian claims like this by noting that, in the second Paralogism, Kant allows that for all we know, the ultimate substratum of matter might be thinking things (A358–360). On Langton's revisionary reading, in the texts quoted in the body of the paper, Kant is asserting that if there were no thinking subjects whatsoever, there would be no matter, because thinking things are the susbstratum of matter. But Langton's revisionary reading is undermined by the fact that Kant only admits the epistemic possibility that matter's substratum is thinking, while in the texts quoted he dogmatically asserts that if there were no thinking subjects, there would be no matter. Whatever Kant means in these passages, he does not mean that the substratum of matter is thinking beings.

  26. 26

    Some readers might wonder what ‘properties an object can appear to have’ means. I intend this expression to refer to what the One Object would reader would call ‘properties an object can have qua appearances.’ I take it that an object can have property F qua appearance only if the object can be experienced as having F. If it is impossible for us to experience an object as F, then the object cannot have F qua appearance. I take it, furthermore, that we do not experience objects as being appearances of things in themselves (in the One Object idiom: as also having properties ‘in themselves,’ or being susceptible of being considered from the ‘transcendental’ point of view). I take it that this is Kant's point at A30/B45 when he writes that the “true correlate” of empirical objects, the thing in itself, “is never asked after in experience.” Thanks to an anonymous reviewer for pressing me on this point.

  27. 27

    In Walker (2010). A similar problem about personal identity is raised in Ameriks (2000), 145.

  28. 28

    Cf. Walker (2010).

  29. 29

    We have to distinguish here between the One Object view as an interpretive claim about Kant, and the view that interpretation attributes to Kant (the appearance and the things in itself are ‘one and the same object’). It is very plausible that in the former sense, as an interpretive claim, the One Object view is at best a rationally warranted hypothesis (as is the Two Object view). But if that hypothesis is true, then it follows that Kant uses ‘appearance’ and ‘thing in itself’ to pick out one and the same object, from which it follows that (according to Kant) we know that ‘appearance’ and ‘thing in itself’ pick out the same object. So, in the argument in the body of the text, the claim is that if the One Object interpretation is true, we know premise (2) in the argument. Thanks to an anonymous referee for pressing me on this point.

  30. 30

    Saying that things in themselves and appearance have different ‘cross-temporal persistence conditions’ might sound odd, because things in themselves are not temporal. However, by this I just mean that from ‘the ship seen at time 1 is the same appearance as time 2’ we cannot infer ‘the thing in itself that appears as the ship at time 1 is not identical to the thing in itself that appears as the ship at time 2.’

  31. 31

    By ‘Monism’ I mean the view about material constitution that is often called the ‘one thing’ view. However, the ‘one thing’ view could be easily confused with the ‘one object’ view; consequently I have decided to call the former view ‘Monism.’ I do not mean to refer to the view, defended in Schaffer (2010), that the whole is ontologically prior to its parts.

  32. 32

    For a survey of views in material constitution see Rea (1997). Influential critics of Monism include Johnston (1992, reprinted in Rea (1997)) and Fine (2003). Monism is defended, in different ways, by Gibbard (1975) and Jubien (1993).

  33. 33

    Gibbard (1975), 96.

  34. 34

    Some Monists may simply deny premise (2) in the argument. I am going to ignore that kind of Monism, because my interest is in developing a One Object reading of Kant that accepts the truth of the corresponding premise in the Modal Indiscernibility argument.

  35. 35

    Some readers might wonder how this strategy compares to Lewisian counterpart theory, as developed, for instance, in Lewis (1968) and (1986). For Lewis, the ‘witnesses’ are going to be, of course, possible worlds. So for Lewis the ‘witnesses’ of (2) are not worlds in which Goliath is smashed and ceases to exist. They are worlds in which a Goliath-counterpart is smashed and ceases to exist. Which objects in which worlds count as Goliath counterparts—thus, which worlds are witnesses to the counterfactual (2)—will depend upon the name used to pick out Goliath. ‘Goliath’ and ‘Lumpl’ will introduce different counterpart relations. So far, I think that the Lewisian counterpart view is isomorphic to the strategy I've outlined. However, Lewis's system introduces the possibility of an additional layer of context sensitivity—the use of a name in one context might determine one counterpart relation, while in another context that name might introduce a different counterpart relation (e.g. the context ‘in the broadest sense of ‘possible’’ might introduce a very wide counterpart relation on which ‘Goliath might have been a poached egg’ will turn out to be true). I'm not sure this additional layer of context sensitivity matters for my purposes, though. Furthermore, I have my doubts as to how Kantian the idea that the truth of these judgments is irreducibly context sensitive could be, aside from the context-sensitivity already encoded in the idea that the empirical names for ‘one and the same’ object function differently. Thanks to an anonymous referee for pointing out this complication to me.

  36. 36

    By saying a relation R is contingent, I mean that if xRy it does not follow that □(xRy). By saying that it is sortal-relative I mean that if it is true that x bears R to y, this is true in virtue of x and y belonging to some sortal K such that x bears R to y qua K and it does not follow that for every other sortal K* x bears R to y qua K*. For more on these issues, see Gibbard (1975) and Geach (1980, reprinted in Rea, 1997).

  37. 37

    I am also assuming that knowledge is closed under known logical entailment.

  38. 38

    A similar problem and similar solution is suggested by Van Cleve (1999), 149–150 (see also n. 43 at 292). Van Cleve generates the problem by assuming that, for Kant, things in themselves obey the identity of indiscernbiles. My argument does not make this assumption.

  39. 39

    Jubien (2009), 86.

  40. 40

    I am using possible worlds here purely as a heuristic point; the same point could be made without them. See Jubien (2009) for a critique of ‘object fixation’ within a theory of modality that makes no use of possible worlds whatsoever.

  41. 41

    Walker (2010).

  42. 42

    I want to make it clear that ‘now’ modifies the object appearing not the way in which the thing in itself appears. Since things in themselves are atemporal, none of their properties can be possessed relative to times. How to understand the relation between atemporal things in themselvs and their temporal appearances lies outside the scope of this paper.

  43. 43

    In the first Critique see the ‘Resolution of the cosmological idea of the totality of division of a given whole in intuition’ (A523/B551-A527/B555). Kant is also especially clear on this point in the Metaphysical Foundations of Natural Science (Ak. 4:503–508).

  44. 44

    I am aware of only one place in Kant's corpus where he addresses something like the conclusion of this argument. In response to Eberhard's mistaken claim that things in themselves are parts of appearances, Kant writes in “On a discovery”: “it is a completely erroneous view of the theory of sensible objects as mere appearances, which must be underlaid by something nonsensible, if we imagine or try to get others to imagine, that what is meant thereby is that the super-sensible substrate of matter will be divided into its monads, just as I divide matter itself” (Ak. 8:210; translation in Kant (2002)). But notice that Kant is denying that thing in themselves are divided in the way that bodies are, i.e. he is denying that they have the same mereological structure. He is not, at least here, denying that we can know that there are at least as many things in themselves as there are sub-regions of space.

  45. 45

    By the ‘same appearance’ I do not mean, of course, numerical identity, but the contingent, sortal-relative identity-like relation described earlier. Giving a complete account of what this relation consists in would require giving an account of what, on Kant's view, grounds the ‘sameness and difference’ of appearances over time and at a time. That lies far outside the scope of this essay. For the sake of my argument, though, it suffices to note that, whatever this relation ultimately consists in, Kant wants to respect our ordinary judgments about the sameness and difference of appearances. So I take it that I can appeal to our judgments about the sameness and difference of appearances in ordinary cases without having to give an account of the nature of this relation. Thanks to an anonymous referee for pressing me to clarify this point.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. 1. Problems for the ‘One Object’ Reading
  5. 2. Monism and Indiscernibility
  6. 3. Monism and the One Object Reading
  7. 4 Why the One Object Reading Nonetheless Fails
  8. References
  • Adams, R.M. (1997). “Things in Themselves.” Philosophy and Phenomenological Research 57 (4): 801825.
  • Adickes, E. (1924). Kant und das Ding an sich. Berlin: Pan Verlag.
  • Allais, L. (2004). “Kant's One World: Interpreting ‘Transcendental Idealism.’ British Journal of the History of Philosophy 12 (4): 655684.
  • Allison, H. (1983). Kant's Transcendental Idealism. New Haven: Yale University Press; second edition, 2004.
  • Ameriks, K. (1982). “Recent Work on Kant's Transcendental Philosophy.” American Philosophical Quarterly 19: 124; reprinted in Ameriks (2003).
  • Ameriks, K. (2000). Kant's Theory of Mind. 2nd edition. New York: Oxford University Press.
  • Ameriks, K. (2003). Interpreting Kant's Three Critiques. New York: Oxford University Press.
  • Aquila, R. (1979). “Things in Themselves and Appearances: Intentionality and Reality in Kant.” Archiv für Geschichte der Philosophie 61: 293307.
  • Bird, G. (1962). Kant's Theory of Knowledge: an outline of one central argument in the Critique of Pure Reason. New York: Humanities Press.
  • Collins, A. (1999). Possible Experience: Understanding Kant's Critique of Pure Reason. Berkeley and Los Angeles: University of California Press.
  • Fine, K. (1982). “Acts, Events and Things.” in Leinfellner, W., Kraemer, E., and Schank, J. (1982).
  • Fine, K. (2003). “The Non-Identity of a Material Thing and its Matter.” Mind 112 (446): 195234.
  • Geach, P. (1980). Reference and generality: An examination of some medieval and modern theories. Ithaca, NY: Cornell University Press; selections reprinted in Rea (1997).
  • Gibbard, Alan. (1975). “Contingent Identity.” Journal of Philosophical Logic 4 (2): 187222; reprinted in Rea (1997).
  • Guyer, P. (1997). Kant and the Claims of Knowledge. Cambridge: Cambridge University Press.
  • Johnston, M. (1992). “Constitution is not Identity.” Mind 101 (401): 89106; reprinted in Rea (1997).
  • Jubien, M. (1993). Ontology, Modality and the Fallacy of Reference. New York: Cambridge University Press.
  • Jubien, M. (2009). Possibility. New York: Oxford University Press.
  • Kant, I. (1900). Kants gesammelte Schriften. Edited by the Berlin-Brandenburg Academy of Sciences. Berlin: Walter de Gruyter.
  • Kant, I. (1998). Critique of Pure Reason. Edited and translated by Paul Guyer and Allen Wood. Cambridge: Cambridge University Press.
  • Kant, I. (2002). Theoretical Philosophy after 1781. Edited by Henry Allison and Peter Heath. Cambridge: Cabmridge University Press.
  • Kripke, S. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press.
  • Langton, R. (1997). Kantian Humility. New York: Oxford University Press.
  • Leinfellner, W., Kraemer, E., and Schank, J. (1982). Sprache und Ontologie: Internationale Wittgenstein Symposium. Vienna: Hölder-Pichler-Tempsky.
  • Lewis, David. (1968). “Counterpart Theory and Quantified Modal Logic.” The Journal of Philosophy 65 (5): 113126.
  • Lewis, David. (1986). On the Plurality of Worlds. Cambridge: Cambridge University Press.
  • McDaniel, Kris. (Ms). “Kantian Ignorance.”
  • Prauss, G. (1974). Kant und das Problem der Dinge and Sich. Bonn: Bouvier.
  • Rea, M. (1997). Material Constitution: A Reader. Lanham, MD: Rowman & Littlefield.
  • Rosefeldt, T. (Ms). “The Identity of Appearances and Things in Themselves.”
  • Schaffer, J. (2010). “Monism: The Priority of the Whole.” The Philosophical Review 119 (1): 3176.
  • Sider, Theodore. (1999). “A Critical Discussion of Michael Jubien's Ontology, Modality, and the Fallacty of Reference.” Noûs 33: 284294.
  • Stang, N. (ms). “Kant's Transcendental Idealism.”
  • Van Cleve, J. (1999). Problems from Kant. New York: Oxford University Press.
  • Walker, R. (2010). “Kant on the Number of Worlds.” British Journal of the History of Philosophy 18 (5): 821843.
  • Westphal, M. (1968). “In Defense of the Thing in Itself.” Kant-Studien 59: 118141.