DENNIS SCHULTING (ed.) | Kantian Nonconceptualism | Palgrave Macmillan, 2016


By Jessica Williams 

Is Kant a conceptualist or a nonconceptualist? Very roughly, this amounts to the following question: Do intuitions depend on concepts in order to represent objects? Much recent Kant scholarship is devoted to answering this question, which is of interest not only for its connection to contemporary debates in philosophy of mind and perception, but also because the answer one provides has important implications for how one understands crucial features of Kant’s account of cognition.

While conceptualism was for some time the default interpretation of Kant, this is no longer the case. In fact, one now finds a number of competing nonconceptualist interpretations of intuition, and conceptualists have in turn divided in their characterisations of the way in which intuitions depend on concepts in order to represent objects. As Lucy Allais notes in her contribution, an important result of this recent debate is “lively dispute and clarification of key terms in Kant’s philosophy, such as intuition, sensation, perception, cognition, and synthesis” (p. 2). The essays in this volume continue in this vein. Engaging with them has certainly helped me to clarify my own understanding of key features of Kant’s account of cognition, including his account of space and spatial representation; his precise strategy in the second-half of the B-Deduction; and the nature and role of intuitions.

In this review, I’d like to focus on the way in which authors in this volume approach the following three questions: (1) Is the unity of the pure intuition of space as an all-encompassing whole independent of the understanding, and if so, what follows with respect to the unity of empirical intuitions? (2) When Kant says that all objects of the senses stand under the categories, does this mean that the categories (and concepts more generally) play a role in generating intuitions, or is there a way of understanding this claim that is compatible with nonconceptualism (according to which intuitions do not depend on concepts in order to be intuitions)? And finally, (3) Are intuitions representational or relational states and what implications does this have for how we understand the role of intuitions in establishing real possibility?

1. The Unity of Space

In the Transcendental Aesthetic, Kant argues that our original representations of space and time are not concepts, but a priori intuitions (A25/B39–40; A32/B48). In §26 of the B-Deduction, however, Kant argues that the unity of space and time “as intuitions themselves” depends on the understanding (B160–1). In a footnote to this passage, Kant distinguishes between mere forms of intuition and formal intuitions, where the latter involve a “unity of […] representation” that “presupposes a synthesis” (B 160–1n). Conceptualists take these remarks to show that the understanding is involved in generating the unity of our pure forms of intuition in virtue of which they count as formal intuitions. At the same time, a number of scholars have argued that the unity of space and time as all-encompassing and infinite wholes cannot be a product of conceptually guided synthesis.[1] In particular, they argue that because synthesis is an activity that always proceeds from part-to-whole to produce representations of determinate spaces and times, it cannot explain the unity of space, which is the unity of an infinite whole that precedes its parts.

In ‘The Difference Between Original, Metaphysical and Geometrical Representations of Space’ (Chapter 11 in the volume), Clinton Tolley offers a helpful taxonomy for discussing the unity of space. The original representation of space is what is given to the subject prior to any determination; metaphysical space is the representation of this space by the metaphysician, who is concerned with articulating what belongs to our concept of the original intuition of space (viz., that it is an infinite, given magnitude); and general concepts of space result from the delimitation of the originally given space and belong to the geometrical representation of space (p. 265). The metaphysical and geometrical representations, while they are derived from the original representation of space, require the understanding. In the case of the geometrical representation of space, the understanding is required because it is via the act of construction that we delimit space. As we shall see in what follows, conceptualists and nonconceptualists alike go astray in failing adequately to distinguish between these different ways of representing space.

This is not taxonomy for taxonomy’s sake; Tolley wants to deploy these distinctions to respond to the conceptualist interpretation of §26, among other passages from the Transcendental Deduction. According to Tolley, the B160–1 footnote is not concerned with the unity of our original representation of space, but rather with the unity that belongs to the representations of this space in geometry (p. 283). The footnote, in other words, does not challenge the claim that the content of intuition is nonconceptual; it merely points out that the conceptual representation of space—when we construct concepts in intuition—depends on the understanding.

More generally, Tolley wants to draw a distinction between that what belongs to intuition (pure and empirical), which does not depend on the understanding, and the representation of that which belongs to intuition, which depends on intellectual conditions of the understanding. As Tolley reads the Transcendental Deduction, Kant is “shifting his target from what is constitutive of a representation (intuition) per se to what is required for the representation of  (certain features of) a representation” (p. 281). That is, in order to become conscious of these features of intuition, the understanding is required.

We ought, however, to separate Tolley’s taxonomy from the use to which he puts it in interpreting the Transcendental Deduction. The problem is that his strategy for accommodating Kant’s arguments in the Deduction does not accord with Kant’s account of geometry.

It seems to follow from Tolley’s distinction (between intuition and what is required for the representation of intuition) that what we are doing when we form geometrical concepts is becoming conscious of the features that are already present in our (original) intuition of space. But, while it is certainly right that the original representation of space is the ground of geometrical constructions, Kant is explicit in the Prolegomena that the laws of geometry are equally grounded in the understanding insofar as it determines space. Consider the following passage, which I quote at length given its importance for my discussion in this, as well as in the next, section:

Here then is nature that rests on laws that the understanding cognizes a priori, and indeed chiefly from universal principles of the determination of space. Now I ask: do these laws of nature lie in space, and does the understanding learn them in that it merely seeks to investigate the wealth of meaning that lies in space, or do they lie in the understanding and the way in which it determines space in accordance with the conditions of the synthetic unity toward which its concepts are one and all directed? Space is something so uniform, and so indeterminate with respect to all specific properties, that certainly no one will look for a stock of natural laws within it. By contrast, that which determines space into the figure of a circle, a cone, or a sphere is the understanding, insofar as it contains the basis for the unity of the construction of these figures. The mere universal form of intuition called space is therefore certainly the substratum of all intuitions determinable upon particular objects, and, admittedly, the condition for the possibility and variety of these intuitions lies in this space; but the unity of the objects is determined solely through the understanding. (Prol, AA 4:321–2).

What I take from this passage is that the role of the understanding in geometry is not a matter of bringing to consciousness a unity that already belongs to intuition per se (to borrow Tolley’s phrase), but of first constituting the relevant unity. Notice that what secures geometrical laws is the fact that “the unity of the objects is determined solely through the understanding”. This point has important implications for Kant’s account of empirical intuition. Even if we grant that the original representation of space has a unity that does not depend on the understanding, it does not follow, as Tolley seems to think, that empirical intuitions have a unity that is independent of the understanding, precisely because determinate spaces (to which the laws of geometry must apply) are the form of empirical intuitions. In their contributions to the volume, Stefanie Grüne and Thomas Land both use this point to argue for moderate forms of conceptualism (Grüne prefers the term “modified Intellectualism”).

In ‘Sensible Synthesis and the Intuition of Space’ (Chapter 4), Grüne begins by reconstructing and evaluating Colin McLear’s mereological argument for the thesis that pure intuitions have a distinct kind of unity that cannot be a product of any act of synthesis. McLear (2015) argues that the unity of the pure intuitions of space and time cannot be a product of any act of synthesis because this unity has whole-part priority and synthesis can only produce representations with part-whole priority. But as Grüne shows, McLear does not disambiguate between two senses of ‘pure intuition’: that which refers to the pure intuitions of space and time as all-encompassing wholes, and that which refers to the pure intuitions of determinate spatial regions and temporal intervals. McLear’s argument, if successful, only establishes the independent unity of pure intuitions in the former sense, but not the latter.

On Grüne’s modified Intellectualist interpretation, the pure intuitions of space and time as singular infinite wholes are independent of the understanding, but the pure intuitions of finite spaces and times—which are the form of empirical intuitions—depend on sensible synthesis. According to Grüne, “the function of sensible synthesis is to make the object of a complex sensible representation phenomenally present to the subject and thereby to generate an intuition” (p. 88). She then argues that it is because the pure intuitions of space and time as infinite wholes are not phenomenally present to the subject that they do not require sensible synthesis. We do not perceive or imagine infinite space; instead, what is phenomenally present to us is a property of finite (i.e. determinate) spaces, “the property of being part of the all-encompassing space” (p. 95).

This strikes me as getting the dependence relation between space and finite spaces the wrong way round. Finite spaces are not given parts of space, but delimitations of an infinite given space. It is hard to see, moreover, how we can arrive at the latter by starting from the perspective of the former. In what sense is the property of belonging to an all-encompassing space (or time) present to us when representing finite spaces? We certainly experience a given region of space as bounded by a larger region of space, but this is not the same as representing it as belonging to a singular, infinite, and all-encompassing space. Grüne is concerned with avoiding the criticism that her modified Intellectualism is ad hoc, but I think a better strategy for responding to this objection is to focus on the requirements for representing determinate magnitudes (rather than with what is and isn’t phenomenally present).

It is precisely consideration of determinate magnitudes that leads Thomas Land, in ‘Moderate Conceptualism and Spatial Representation’ (Chapter 7), to argue that Kant’s theory of spatial representation implies a moderate form of conceptualism. For Land, the function of sensible synthesis is to generate intuitions (both pure and empirical) of determinate magnitudes. To support this interpretation, he appeals to the argument in the Axioms of Intuition that the “same synthesis” that generates the representation of determinate magnitudes is involved in the apprehension of appearances in empirical intuition (B202–3). This is only “moderate” conceptualism, because although sensible synthesis is an act of spontaneity, it is not an act of judgement. Two main features of sensible synthesis distinguish it from judgement: (1) unlike judgment, sensible synthesis involves a successive synthesis of a manifold; and (2) unlike judgement (which has a predicative structure) sensible synthesis has a spatio-temporal structure.

I think Land is right to distinguish between sensible synthesis and judgement along these lines, and also right that sensible synthesis is involved in the representation of determinate magnitudes in empirical intuition. But this is not the only, or even main, point of his contribution. He is primarily concerned with responding to the objection that moderate conceptualism cannot account for the holistic character of spatial representation, but is instead committed to “sense-atomism”, the view that synthesis produces the representation of objects out of atomic sensations that otherwise lack determinate relation to each other. Land claims, however, that although sensible synthesis proceeds from part to whole, this is compatible with “the holistic character of spatial representation” (p. 168). The holistic unity of space is a feature that arises from the nature of space as a pure form of intuition, but only as a potentiality. This potentiality must be actualised through the a priori determination of sensibility by the productive imagination (through the act of figurative synthesis). Because this a priori actualisation of sensibility is a condition of empirical intuitions, “sensations will be just as holistic as the spatiotemporal form of the capacity whose actualisations they are” (p. 162).

But, as Land himself argues, the figurative synthesis that is a condition for the apprehension of an object in empirical intuition is the synthesis that produces the representation of a determinate magnitude. The problem is that this kind of synthesis cannot explain the unity that belongs to space and time as infinite and all-encompassing wholes. Or, at the least, Land would need to explain what the relationship is between the kind of figurative synthesis that produces the representation of determinate magnitudes in geometry (and in empirical intuition) and the act of the productive imagination that actualises the representation of space and time themselves as infinite wholes.

Despite this worry, an important point that Land makes with respect to the discussion of actualisation is that although the understanding (via the imagination) plays a role in actualising sensible representations, this does not mean that the content of these representations is conceptual. For example, we need the understanding in order to represent empirical intuitions as determinate magnitudes (magnitudes that are built up from their parts), but because these intuitions are sensible representations that inherit features from the pure forms of sensibility, they are also continuous magnitudes. That is, we need not be committed to the claim that sensibility operates independently of the understanding in order to acknowledge that sensibility contributes distinct representational content to cognition.

Robert Hanna would likely challenge that latter claim. In ‘Directions in Space, Nonconceptual Form and the Foundations of Transcendental Idealism’ (Chapter 5), Hanna appeals to Kant’s discussion of incongruent counterparts (which traces to the 1768 ‘Directions in Space’ essay) to argue for a non-intellectualist version of transcendental idealism, according to which appearances necessarily conform to space and time, but not to the categories. According to Hanna, incongruent counterparts, especially as perceived by non-rational animals, are “rogue objects”, that is, objects that necessarily fall outside the categories. Because Kant’s goal in the B-Deduction is to show that “the categories are necessarily applicable to all appearances,” the existence of such rogue objects shows that Kant’s argument is unsound (p. 114).

As I see it, however, there is a difference between attributing to non-rational animals a primitive feeling of the difference between the sides of their bodies and attributing to them the ability to cognise incongruent counterparts. Furthermore, the existence of the former ability is only incompatible with Kant’s argument in the Transcendental Deduction if one assumes a strongly conceptualist interpretation of the Deduction, that is, if one takes Kant to argue that the categories are conditions of appearances themselves. If we reject that interpretation, as other authors in the volume suggest, then nonconceptualism need not threaten the goal of the Transcendental Deduction.

2. Nonconceptualism and the Transcendental Deduction

The B-Deduction is usually considered to provide the strongest support for the conceptualist interpretation of intuitions. It is here, after all, that Kant writes, “all synthesis, through which perception itself becomes possible, stands under the categories” (B161). Many conceptualists have taken this to mean that the categories are involved in the production or generation of intuitions themselves, and moreover, that it is because the categories play this role that we can establish a priori that the categories necessarily apply to any object of intuition. Unlike Hanna, most nonconceptualists do not want to accept this characterisation of Kant’s argument. In this section, I evaluate some of the different strategies presented in this volume for reconciling nonconceptualism with the Transcendental Deduction.

The strategy employed by Dennis Schulting, Lucy Allais, and, as we have already seen, Clinton Tolley, is to argue that the categories are not constitutive of intuitions—they do not produce or generate them—instead, they are epistemic conditions of intuition (Schulting, p 254). That is, in order to represent or cognise that intuitions have certain features or properties, we must employ the categories, but prior to this employment, intuitions already have these features.

In ‘On an Older Dispute: Hegel, Pippin, and the Separability of Concept and Intuition in Kant’ (Chapter 10), Schulting employs this distinction to criticise Robert Pippin’s Hegelian interpretation of the Deduction. Hegel famously interpreted the B-Deduction as the place where Kant himself paved the way for overcoming the duality of sensibility and understanding. On Pippin’s reading of the Transcendental Deduction, Kant overcomes the duality of the faculties by arguing that the categories are conditions of anything being given in sensibility at all. Schulting grants to Pippin that the categories are conditions for anything being given, but only in an epistemic sense. That is, he grants that the thinking subject must apply the categories to sensible content in order to grasp this content as the sensible content of thought. But, for Schulting, it is a mistake to conclude from this that the categories themselves constitute or generate this content. Because the categories are only epistemic, and not constitutive, conditions of intuition, we can maintain Kant’s distinction between the faculties of sensibility and understanding, rather than taking Kant to collapse this distinction in the B-Deduction.

The focus on epistemic conditions is likewise central to Lucy Allais’ strategy for reconciling nonconceptualism with the Transcendental Deduction.[2] In ‘Conceptualism and Nonconceptualism in Kant: A Survey of the Recent Debate’ (Chapter 1), she writes:

[T]he argument of TD is not concerned with what is necessary for intuitions to be intuitions (to be unified in the way that is necessary for them to be singular and immediate representations that give us objects), but for intuitions to be unified in the different way that is necessary for them to be cognized—for them to be grasped as objects using concepts and to have concepts applied to them. (p. 19–20).

For Allais, empirical intuitions already present us with sensible particulars prior to any conceptual determination. Elsewhere, she defines a particular as a “spatially continuous and unified individual[] existing outside the subject and located in space” (2009:405). In order to grasp these particulars as objects, that is, as unified complexes of properties, we must employ the categories.

I share Sacha Golob’s concern with this strategy. The problem is that this kind of move renders Kant’s argument vulnerable to the criticism that it only shows that the subject must apply the categories in order to think about objects, but not that the categories must apply to these objects. Golob cites Anil Gomes (2010, 2014), who argues that to avoid this criticism, we must read Kant’s argument in the second-half of the B-Deduction as showing that the categories are not just conditions on our thought about objects, but play some role in the constitution of these objects.

In order to meet this constraint, Golob suggests a different strategy. In ‘Why the Transcendental Deduction is Compatible with Nonconceptualism’ (Chapter 2), Golob argues that the categories are “a necessary condition on and only on the representation of a certain privileged class of spatial or temporal relations” (p. 38). He has two kinds of relation in mind: (1) the representation of composition (i.e. representing something as a whole that is built up out of parts) and (2) the representation of the distinction between subjective and objective succession.

On this view, the categories are not just conditions for making judgements, but also conditions for certain kinds of sensible representation of objects. This is still a version of nonconceptualism, because categorial synthesis is not a condition for having empirical intuitions (understood as representations of spatiotemporal particulars), but only for sensibly representing certain complex relations among intuitions (e.g. relations of composition or objective succession).

While this is a promising strategy, Golob’s version of it may collapse into the previous strategy (in particular, Lucy Allais’ version as presented by Golob). Golob criticises Allais, because for her, the categories are just conditions for the application of concepts to objects in judgement, and he thinks this cannot accommodate the second-step of the B-Deduction. In order for his view to be different, it must be the case that the categories are a condition for the sensible representation themselves. But if that is so, then it won’t be the case that a non-human animal that lacks the categories can sensibly represent—in this case, see—these relations. Yet Golob claims that on Kant’s view, “an animal can indeed see a line at a glance” but it cannot “represent mereological relations between the whole and its parts” (p. 41). But, if that is the case, then it sounds like the latter representation is simply the judgement that the line (segment) is a whole built up out of its parts.[3]

It is worth thinking about why Kant thinks that synthesis is required to represent relations of composition and succession. At the beginning of his discussion of the synthesis of apprehension in the A-Deduction, Kant writes:

Wherever our representations may arise, whether through the influence of external things or as the effect of inner causes, whether they have originated a priori or empirically as appearances—as modifications of the mind they nevertheless belong to inner sense, and as such all of our cognitions are in the end subjected to the formal condition of inner sense, namely time, as that in which they must all be ordered, connected, and brought into relations. (A98–9)

The thought here is that in apprehension we do not just take in the relations that were already present in intuition prior to any synthesis, but instead, we introduce the very relations into intuition insofar as we connect the manifold in inner sense.

Here, we arrive at the crux of the debate between conceptualists and nonconceptualists. Conceptualists think that synthesis is required to generate the unity of the manifold in virtue of which intuitions represent objects to the perceiving subject at all. They would deny that on Kant’s account the animal sees a line but cannot represent it as such. The animal, after all, might be given something we might want to call parts of the line, but if the animal cannot reproduce the parts of the line in its inner sense, then it does not see the line.

In fact, this is not just a matter of reproduction. Lines, for Kant, are produced via the motion of a point (that is, through the subject’s description of space). I think that Kant could grant that a non-human animal can detect things like edges, but this is not the same as seeing a line. This brings me back to the long passage from the Prolegomena that I cited in the previous section. Kant thinks that the lawfulness of geometry is secured because the unity of spatial constructions is a result “solely” of the understanding. And, I take it, what secures the applicability of the laws of geometry to empirical objects is that in the course of empirical intuition, we delimit the (pure) spatial form of objects. But does this mean that the understanding generates intuitions through the act of synthesis that delimits or determines the spatial (or spatio-temporal) form of empirical objects? How do we understand the role of the productive imagination in empirical intuition?

This is where things get tricky. Conceptualists, even of the moderate variety, slide from the claim that the understanding (via the productive imagination) generates the unity of intuitions to the claim that the understanding / imagination generates intuitions (see Land, ch. 7, p. 157). This is perhaps not a problem when what we are talking about is the construction of intuitions in geometry, but what about in empirical intuition? Notice that in one of the passages where Kant explicitly turns his attention to the role of the understanding (productive imagination) in empirical intuition—the house example from §26 of the B-Deduction—he doesn’t claim that we draw the shape of a house, but that we “as it were draw its shape” (B162). So Kant is not exactly identifying what we do in empirical intuition with what we do in the course of pure intuition (where we do draw a figure). This passage does not necessarily support the claim that synthesis produces or generates intuitions.

Nonconceptualists, in other words, can read Kant as here claiming that although the shape is already presented to us in empirical intuition, we must synthesise what is presented in order to cognise this shape.[4] Again, as we have already seen, this is what it means to claim that the understanding is an epistemic, not a constitutive condition of intuition. Such an interpretation is not without textual and philosophical support. As Lucy Allais emphasises, Kant repeatedly characterises the role of intuition as that of giving us objects (A19/B33,  A50/B74, A89/B122; Allais 2009, 2015). And as Kant notes in §13 of the Deduction, the categories, unlike the forms of space and time, “do not represent to us the conditions under which objects are given in intuition at all” (A89/B122).

I think that nonconceptualists are right to resist the claim that the understanding and the imagination generate or produce empirical intuitions. But—and this is important—it is a mistake to think that they can help themselves to what is given in intuition prior to synthesis. Christian Onof nicely diagnoses the mistake in his contribution to the volume, ‘Is There Room for Nonconceptual Content in Kant’s Critical Philosophy?’ (Chapter 9). Onof writes:

What nonconceptualists are in effect doing is implicitly relying upon the fact that objects of experience are in place to consider some other way of accessing these same objects. Such a move could be justified on an interpretation of Kant’s metaphysics which ignores the central Critical move according to which fully determinate objects are not available to be picked out, but cognition of objects involves their being determined in conformity with the transcendental conditions of experience. (p. 218)

What I especially like about this way of putting the worry is that it also gives us a way of thinking about what the understanding/imagination is doing that is more than an epistemic condition of intuition, but which does not collapse the distinction between sensibility and understanding by claiming that the understanding generates or produces (empirical) intuitions. It seems to me that one of the main roles of the productive imagination in empirical intuition is to delimit and determine the boundaries of an object (and that this is what Kant means when he says that we “as it were” draw shapes in empirical intuition). There is simply no determinate fact of the matter about what shape the “object” has before synthesis, because no object has been differentiated. Subtracting from the work of the understanding (imagination) in this respect is to smuggle in, as Onof puts it, a “residual Strawsonian realism” (p. 219). We know that the laws of geometry will apply to empirical objects because the same rules of synthesis to which we appeal in geometrical constructions govern the delimitation of the spatial form of objects.

Although he does not think that empirical intuition presents a determinate object prior to synthesis, Onof nevertheless thinks that prior to synthesis “we are conscious of particulars defined through patterns of sameness and difference in space, without being conscious of them as distinct” (p. 220). Onof characterises this as the perceptual cognition of things, or PCT for short. This is not cognition in the normal sense (which requires conceptually-guided synthesis), but a weaker form of ‘cognition’ that results from the mere apprehension and reproduction of the manifold as guided by empirical laws of association. A creature that enjoyed PCT could thus make distinctions in navigating its environment by responding to these patterns of sameness and difference, without being conscious of particulars as distinct. Onof appeals to PCT not only to explain animal consciousness (and even aspects of human consciousness), but also thinks that it has an important role to play in explaining the way that inclinations can influence human behaviour in the practical sphere without our awareness of them. It can thus explain why we can never be sure that we are acting out of duty. Even if we have ruled out the influence of inclinations of which we are aware, there may always be inclinations that unconsciously influence us.

One point that comes out of Onof’s discussion is that if we want to isolate a contribution of sensibility that is wholly independent of the understanding, it can at most play a subpersonal role. It may be useful to do so, for example, in the kinds of contexts that Onof has in mind—explaining non-human animal behaviour and even some human behaviour—but it is worth noting that the nonconceptual content that Kant is primarily interested in is that which must be available to the judging subject. For example, the synthetic judgements of mathematics are grounded in constructions in intuition. These constructions make spatial content available that is not ‘contained’ in the concepts. This content is and must be available to the judging subject, but it is also content that is only available insofar as she has constructed the concept in intuition. Thus, it is not independent of the understanding. But the role of the understanding in ‘actualising’ this content—to borrow Land’s phrase—does not render the content conceptual (or at least not fully so).

There is another domain in which nonconceptual content must be available to the judging subject, namely, in aesthetic judgements. In his contribution to the volume, ‘Kant’s Aesthetic Nonconceptualism’ (Chapter 6), Dietmar Heidemann appeals to Kant’s account of aesthetic judgements in the Third Critique to argue that Kant is a nonconceptualist. This is because the determining ground of a judgement of taste is nonconceptual, namely, an aesthetic feeling. This feeling is not just a nonconceptual state, according to Heidemann, but has nonconceptual representational content, where what it represents is “the harmonious relation of the understanding and imagination” (p. 130). There is, of course, a debate in the scholarly literature concerning whether aesthetic feeling is representational, and I won’t weigh in on that debate here.

It is worth noting, however, that even if we think that judgements of beauty depend on nonconceptual feelings that represent the state of mind of the subject, this is not the kind of nonconceptual content at stake in the recent debate, which concerns whether intuitions have nonconceptual content. It might be worth thinking, however, not just about the nonconceptual aesthetic feeling involved in aesthetic judgements, but also about the kind of intuitive content that gives rise to the harmonious free play of the faculties on which these judgements are grounded.

So where does all this leave us with respect to the question of whether nonconceptualism is compatible with the Transcendental Deduction? This will obviously depend on how one construes nonconceptualism and how one construes Kant’s goal in the ‘second step’ of the B-Deduction. But I do think there is at least one way of reading the Transcendental Deduction that acknowledges both that the categories do not generate or produce empirical intuitions and that they are more than merely epistemic conditions of intuition (because they—in particular the mathematical categories—first constitute any determinate relation to an object in empirical intuition by first delimiting the boundaries of the object).

3. The Nature and Role of Intuitions

Throughout this review, I have been talking about the content of intuitions. This is because much of the debate between conceptualist and nonconceptualist interpreters of Kant has been framed in terms of the representational content of intuition. That is, the central question was whether the representational content of intuition was conceptual or nonconceptual.

Colin McLear, in a recent paper (2016) and in his contribution to this volume, ‘Getting Acquainted with Kant’ (Chapter 8), however, denies that intuitions have content (of either the conceptual or nonconceptual variety). To say that a state has content is to say that it has correctness conditions, that it represents the world correctly or incorrectly. For McLear, intuitions are not representational states, instead, and in line with Allais (2015), he suggests that intuitions are relational states, where a relational perceptual state directly presents the environment to the perceiver (such that the objects in one’s environment partially constitute the state).

To get at the difference between these views, consider your perception of one of the books on your desk. If I were to swap it with an identical copy, a representationalist (leaving aside those who subscribe to object-dependent variants of the view) would say that your perceptual state is the same, because it has the same content, but a relationist would deny this. Although the two states are subjectively indistinguishable, they are nevertheless different, because in each state you perceive a different book (this follows from the claim that the objects themselves partially constitute the state). Notice that for the relationist, there is no intermediary (e.g. a representation) between you and the objects you perceive. The perceptual state in question thus cannot be characterised in terms of correctness conditions. Through this state one is simply aware of one’s environment.

So why think that intuitions are relational states? They cannot be representational, according to Mclear, because Kant denies that intuitions have correctness conditions. Here, McLear has in mind Kant’s claim that “truth and illusion are not in the object insofar as it is intuited, but in the judgment about it insofar as it is thought” (A293/B350; McLear 2016). One way to read this passage is that Kant is simply reiterating the early modern view of judgement as the act of affirming or rejecting some content (what we would now call a proposition). But, as McLear notes, Kant also uses the term ‘judgement’ to refer to the cognitive act whereby we first generate judgmental content. Synthesis, for McLear, is precisely the act through which concepts “are brought together to form truth-bearing contents” (p. 174) in judgements. For the relationist, it is thus at the level of judgements that we have representational content.

The biggest problem for this view is that it is not well-suited to explain pure intuitions or to explain the connection between pure and empirical intuitions that secures the application of geometry to empirical objects. Pure intuitions are not relational states.  This is true of the pure forms of intuition (as all-encompassing wholes, which are not objects for Kant) as well as of the pure intuitions of determinate spaces and times in geometry. In the latter case, the relationist could respond by saying that when we are doing geometry, we are representing the form of empirical objects, while in empirical intuition, we are simply presented with the form of objects (along with their other features). But I do not think this accords with Kant’s argument in the Axioms of Intuition. According to Kant, what secures the applicability of geometry to empirical objects is that “the synthesis of spaces and times, as the essential form of all intuition, is that which at the same time makes possible the apprehension of the appearance” (A166/B207). Another way to put the worry is that, as Land argues, sensible synthesis (what Kant is here referring to as “the synthesis of spaces and times”) while it may involve judgements, is distinct from the act of judgement. In geometry, for example, the result of sensible synthesis is an intuition. And, as I have already suggested, although we do not produce empirical intuitions, it is via sensible synthesis that we delimit the spatial form of objects. It is not clear, in other words, how the relationist can accommodate sensible synthesis.

A further dimension of the debate between relationists and representationalists concerns the role of intuitions in establishing real possibility and thus satisfying the modal condition on cognition. The modal condition is as follows:

Necessarily, S cognizes an object O only if S is in a position to prove the real possibility of an object possessing the features constituting the content of the representation of O. (McLear, p. 189)

The purported advantage of relationism is that because intuitions are object-dependent, they can prove the real possibility of objects by way of the actuality of these objects (since objects are partial constituents of the intuition). Of course, because some versions of representationalism are also object-dependent, the real distinction here is between object-dependent and object-independent interpretations of intuition, which are the subject of Anil Gomes and Andrew Stephenson’s contribution ‘On the Relation of Intuition to Cognition’ (Chapter 3). Object-Dependent (OD) views hold that “necessarily, if there exists an intuition i of some object, o, then o exists”, while Object-Independent (OID) views deny this (p. 55).

Gomes and Stephenson endorse the modal condition on cognition and, crucially, they assume that “bringing an intuition under a concept suffices for cognition” (p. 61). From this, it follows that the OID and OD views offer different accounts of the nature of cognition and the role that intuition plays in proving real possibility. The set of objects one can cognise on the OID view is larger, and so, too, is the set of really possible objects. The explanation for this is that on the OID view, formal possibility is either identical to or entails real possibility. Since a hallucination of an object accords with the formal conditions of experience (the categories and space and time), it suffices to prove real possibility. Indeed, only the OID view can account for the real possibility of non-actual objects (p. 69). The set of things one can cognise on an OD view is obviously more restricted. Furthermore, as we have seen with McLear’s view, (empirical) intuitions put one in a position to prove real possibility by way of actuality. The OD view, as Gomes and Stephenson note, collapses the “(extensional) distinction between the really possible and the actual” (p. 71).

I do not have settled views about this debate (and, I should note that Gomes and Stephenson are primarily interested in laying out the implications of each view of intuition, rather than with defending a particular view). But I do have two sketchy worries.

First, it is not clear to me that the OID view has to take (most) hallucinations as establishing real possibility. Keep in mind that for Gomes and Stephenson, “bringing an intuition under a concept suffices for cognition” (p. 61). But to bring an intuition under a concept is to make a judgement. And a hallucination as of a pink elephant need not lead to the judgement “That is a pink elephant”. In other words, that the concepts <pink> and <elephant> occur to me in the course of my hallucination does not mean that I subsume this representation under the concept <elephant>, although I might. Perhaps a better example is the hallucination as of a flying elephant. It is precisely because I know that elephants cannot fly that I’m not going to make the judgement “That is a flying elephant”.

Another way to put this point is that when we subsume intuitions under concepts, we do so against the backdrop of our understanding of the system of concepts and empirical laws that are in place. The cognition (judgement) might still be false, for example, if, I mistakenly apply the concept <gold> to fool’s gold. We can thus maintain the distinction between cognition and knowledge, while recognising that there is more to cognition than just applying any old concept to any old intuition.

McLear cites a passage in which Kant claims that the possibility of empirical concepts must be “grounded in experience and its known laws” (A123/B270; McLear, p. 191). While McLear is certainly right that more than formal possibility is required to establish the real possibility of empirical concepts, what is doing the work here is experience as comprehensively systematised in empirical laws, not the object-dependence of intuition. Furthermore, since the relationist admits that a relational perceptual state is subjectively indistinguishable from a hallucination, she is in the same position as the representationalist. That is, she cannot appeal to her perceptual state in order to determine whether she is having an intuition or merely hallucinating; she can only appeal to how well this state coheres with what she knows about how the empirical world in fact works. Again, elephants cannot fly, so if I have a (re)presentation of a flying elephant, I judge that it is a hallucination.

4. Some Final Remarks

The debate between conceptualist and nonconceptualist interpreters of Kant has loomed large in recent years, and will undoubtedly continue to do so. One reason for this is that although this way of framing the debate—“Is Kant a conceptualist or nonconceptualist?”—is relatively new, the underlying question is not. This is the question of how to make sense of the distinction between sensibility and understanding that is at the heart of Kant’s Critical philosophy. Kant is committed both to the claim that sensibility and understanding are different faculties with different roles to play in cognition and to the claim that these faculties must work together to produce cognition. Although the debate between conceptualist and nonconceptualist interpreters may not be settled anytime soon, engaging with it promises increased insight into Kant’s Critical philosophy, a point to which the essays in this volume attest.


[1] See for example Messina (2014), McLear (2015), and Onof & Schulting (2015).

[2] An important difference between Schulting and Allais is that Schulting denies that intuitions secure reference to particulars prior to any involvement of the categories (see Schulting, p. 232n.7).

[3] Golob draws a distinction between nonconceptualism about animals (NCA) and nonconceptualism about humans (NCH). But I do not think this distinction helps on this point. If we grant that concept-guided synthesis is required to introduce certain relations into intuition in the first place, then we must deny that any creature that cannot synthesise in accordance with concepts is presented with these relations on the Kantian account.

[4] They thus want to draw a distinction between empirical intuition, which is prior to any act of synthesis, and perception, which is the product of synthesising what intuition (re)presents in such a way that it is available for cognition. See Allais (Chapter 1, p. 6) and Golob (Chapter 2, p. 51) for discussion of this distinction.



Allais, L. (2009), ‘Kant, Nonconceptual Content and the Representation of Space’, Journal of the History of Philosophy 47(3): 383–413.

——— (2015), Manifest Reality: Kant’s Idealism and his Realism (Oxford: Oxford University Press).

Gomes, A. (2010), ‘Is Kant’s Transcendental Deduction Fit For Purpose?’, Kantian Review 15(2): 118–37.

——— (2014), ‘Kant on Perception: Naïve Realism, Nonconceptualism, and the B-Deduction’, Philosophical Quarterly 64 (2014): 1–19.

McLear, C. (2015), ‘Two Kinds of Unity in the Critique of Pure Reason, Journal of the History of Philosophy 53(1): 79–110.

——— (2016), ‘Kant on Perceptual Content’, Mind 125(497): 95–144.

Messina, J. (2014), ‘Kant on the Unity of Space and the Synthetic Unity of Apperception’, Kant-Studien 105(1): 5–40.

Onof, C. & D. Schulting (2015), ‘Space as Form of Intuition and as Formal Intuition: On the Note to B 160 in Kant’s Critique of Pure Reason’Philosophical Review 124(1): 1–58.

© Jessica Williams, 2018.

Jessica Williams is Assistant Professor of Philosophy at the University of South Florida. Williams obtained her Ph.D in Philosophy from Stanford University in 2017. She specialises in Kant and the philosophy of perception and has among other places published in the History of Philosophy Quarterly and British Journal of the History of Philosophy