DENNIS SCHULTING | Kant’s Radical Subjectivism. Perspectives on the Transcendental Deduction | Palgrave Macmillan 2017


By Sacha Golob

In his new book, Kant’s Radical Subjectivism, Schulting provides a rigorous and persuasive account of the core themes of the Transcendental Deduction. I have learnt a great deal from this work, and I am sympathetic to many of its points. In this response, however, I think it will be most interesting to concentrate on two issues where Schulting and I disagree, and where that disagreement has important structural consequences. The first issue concerns the role of objectivity in Kant’s argument, the second the prospects for nonconceptualism. I shall begin by summarising Schulting’s stance on each. I will then explain where we differ and why it matters.

1. Overview of Schulting on Objectivity and Nonconceptualism

Schulting holds that the Transcendental Deduction (henceforth simply ‘the Deduction’ or ‘TD’ in Schulting’s own text) identifies a series of necessary and sufficient conditions for representing something as an object:

The general conception of objectivity at stake in TD is […] how, given the constraints of discursive cognition, objects can be veridically represented at all, and so how empirical objects are constituted qua their being objects. (p. 73)

The key idea here is that of being ‘constituted as’ or ‘qua’ an object:

The “as” here is important […] There is an object only insofar as my representations are taken as constituting an object. (p. 105)

This is a familiar Kantian move: to represent something as an object, where that goes beyond merely having a succession of mental states with some common cause, requires synthetic activity, or constitution, on my part. The underlying model is one on which

we have nothing to go on but manifolds of representations and cannot, as it were, go beyond our representation to point, as it were, to the object. (p. 170)

By extension, any notion of objectivity more sophisticated than the unknown thing-in-itself must be generated through synthesis, specifically categorical synthesis. To put the point in modern terms:

Categories are transcendental conditions of objective experience, they are by definition designators of object-intentionality or objectivity. (p. 154)

As Schulting stresses, Kant analyses this categorical synthesis in part by aligning it with the unity of apperception. The result is a key part of the ‘Subjectivism’ of the book’s title. This is the first of four claims Schulting makes in cashing that term (I return to this list below):

Kantian subjectivity, in virtue of the principle of apperception or transcendental self-consciousness, is solely constitutive of the very conception of what an object is or what objectivity means, so that objective validity must be seen as intrinsic to thought itself. (p. 22)

Schulting is thus naturally opposed to anyone he takes to be assuming, rather than constructing, objective content: such authors, for example Stephenson, are “plain old realist[s], who assume the existence of objects as distinct from, and outside of, our experience of them” (p. 83).

What of nonconceptualism? ‘Nonconceptualism’ of course means very different things across the various philosophical literatures. In a Kantian context, it is standardly understood as the thesis that a subject may possess empirical intuitions of spatiotemporal particulars even if that subject entirely lacks conceptual capacities; this, for example, is the view identified by Allais in her seminal 2009 article (Allais 2009:384). The subsequent debate has often coalesced around the status of non-rational animals (henceforth simply ‘animals’) because these are naturally suited to the role of such subjects: I am not going to rehearse that literature here, but nonconceptualists typically also hold that Kant’s scattered remarks on such animals show that he did indeed see them in that way.

In his book, Schulting makes two main claims regarding nonconceptualism. The first is a textual one: he argues, contra authors such as Hanna (e.g. Hanna 208:62), that it is a mistake to identify ‘figurative synthesis’ as a form of nonconceptual combination (p. 217). I agree with this: given the various inter-definitions between transcendental imagination, synthesis speciosa and the understanding, it is simply a mistake for the nonconceptualist to make this exegetical move.[1] Clearly, everyone needs to allow animals some forms of combination—to take an obvious example, Kant talks explicitly of animals joining representations by association (Br, AA 11:52)—but I think Schulting is absolutely right to separate off ‘figurative synthesis’ as that phrase is used in the Deduction.

Second, Schulting aims to offer something of a compromise on the philosophical issues. On the one hand, he accepts that the nonconceptualist is correct that unconceptualised representations do indeed exist. As he puts it:

There are intuitions that not only are not conceptualised de facto but also need not or indeed could not be conceptualised […]. Kant is not a conceptualist in the sense of the thesis that even to have representations, or indeed intuitions, already involves the categories. (pp. 240, 260)

A blanket conceptualism thus overstates the role which the understanding plays in modelling subjectivity (p. 239). On the other hand, however, the conceptualist is right that the categories are very basic indeed. For example, and unsurprisingly given his views on objectivity, Schulting sees them as necessary to secure reference to objects:

[Nonconceptualists] are wrong to claim […] that these intuitions are objectively valid cognitions or refer to objects. (p. 197)

Indeed, without the synthesis of the understanding, intuition cannot even generate a perception of “determinate spaces” (pp. 22, 275). This is listed as the second of Schulting’s four foundational principles:

Kantian subjectivity, in virtue of the principle of apperception, given sensory input, is solely constitutive of the possibility of perceiving objects as determinate spaces, without however thereby reducing space as infinite given magnitude to being the product of the understanding. (p. 22)

I think Schulting’s final clause is right in rejecting the kind of intellectualism that would effectively collapse the distinction between the Aesthetic and Analytic. The problem, I shall suggest, lies with the rest of Schulting’s principle, and with the close links he postulates between ‘determinate spaces’ and apperception.

2. An Underlying Problem for Schulting: Nonconceptualism as a Case Study

Schulting’s overall position is a complex one and there is a huge amount that I have had to leave out: for example, the impressive subtlety with which he handles the relationship between the categories and apperception (see especially Chapter 4). But I want to keep the focus tight and so it is time to turn to where we disagree. I’ll present my case in two stages. In this section, I shall try to identify an underlying problem with Schulting’s account: I shall use the case of nonconceptualism to show its consequences. In Section 3, I then try to get to the root of the difficulty.

As sketched, Schulting reads the Deduction as an account of the conditions on the representation of objectivity. I am in complete agreement with this: it seems clear to me that Kant is not primarily interested in epistemology, but in intentionality and content. Questions about what we can know about objects are secondary to the question of how, in an idiom beloved by both Schulting and the phenomenological tradition, we even encounter something ‘as’ an object. But this raises a natural question: what is meant by ‘objectivity’? After all, it is clearly a heavily theory-dependent term: what one counts as an ‘object’ for Kant is very different than for Russell or Frege. Schulting himself is well-aware of this and frequently reproaches nonconceptualists for confusing different senses of objectivity:

Kantian nonconceptualists do not seem to differentiate between the various senses of ‘object’ in Kant’s texts. (p. 21)

More specifically, he thinks they fail to distinguish between Gegenstand and Objekt: this leads them to mistake the mere fact that something affects sensitivity, a Gegenstand, with a full blown “objectively valid relation to objects”, an Objekt (p. 21). Schulting’s own position on nonconceptualism follows directly from this move: his view is effectively that animals can have Gegenstände but not Objekte. By extension, they are unable to have an experience of “distinct particulars (Objekte) in bounded space” (p. 305).

Now, I completely agree that Kant’s position is structured around the conditions on objective representation. But I think that the argument is more complex than Schulting allows. In particular, I think that, over the course of the Deduction and the Principles, Kant moves back and forth between several distinct and non-coextensive definitions of objectivity—I look at some detailed examples in Section 3 below. My worry with regard to Schulting is that he sticks to the Gegenstand/Objekt distinction too rigidly with the consequence that he tends to operate with only a binary model of objecthood. In contrast, like Longuenesse (1998:111), I regard that terminology as too inconsistently used by Kant and too crude to really track what is going on. We need, of course, to be cautious in departing from Kant’s original wording, but that must be open when the distinction is employed as haphazardly as this one and when doing so can make better sense of the overall argumentative arc of the text. After all, neither conceptualism nor nonconceptualism can accept at face value everything Kant says in every text.

I want now to illustrate some problematic consequences of Schulting’s binary approach with respect to nonconceptualism.

First, it fails to engage with what was really at stake in the existing debate. Both conceptualists and nonconceptualists agree that what is at issue is the possibility of nonconceptual intentionality. There are, of course, multiple competing theories of intentionality, on both the relationalist and representationalist ends of the spectrum, but the standard tactic in the Kant literature has been to cash intentionality in terms of object-directedness. Consider this summary of the debate by Ginsborg, a leading conceptualist:

That debate, as Allais helpfully puts it, is about the possibility of intentional content without concepts: whether we can have nonconceptual representations which are object-directed, or which represent objects to us. (Ginsborg 2008:68)

The problem is that Schulting’s compromise doesn’t mark a real compromise since it explicitly denies objective content in the absence of conceptual capacities: in other words, his compromise doesn’t touch what all sides agree is key to the debate. To put it another way, Allais defined nonconceptualism in contrast to what she rightly called the “dominant” view as illustrated by Falkenstein’s insistence that that “for Kant, our senses are insufficient for the perception of particular objects” (2006:141). But this dominant view is equally a statement of Schulting’s own position. What Schulting offers is, at least from the nonconceptualist side, not much of a compromise at all: it is just another conceptualism even if it does avoid the most extreme versions of that doctrine.

Perhaps that is simply so much the worse for such nonconceptualism? Well, that brings me to my second point. Animals are prima facie capable of perceiving spatio-temporal particulars standing in quite specific relations: dogs can easily be trained to only take food when it is in one specific position, such as in a bowl, and to leave it when it is next to the bowl or on the table (or even on the end of the dog’s nose as countless YouTube videos attest). Some animals are prima facie capable of tracking far finer-grained spatio-temporal relations than we are: consider an eagle correcting its flight path to time its strike as the prey runs diagonally along a sloping terrain. Such examples led Tyler Burge (2010:23) to describe the view that animals cannot perceive spatio-temporal particulars as simply “empirically refuted”.

Schulting at times notes this. He observes, for example, that an elephant calf squeezing through a small whole in the fence, rather than going round to the gate as its mother does, must have some awareness of its own width in relation to the gap (p. 308). Yet it is hard to see how Schulting’s position can accommodate that reality. Recall that the second key thesis of the book states that:

Kantian subjectivity, in virtue of the principle of apperception, given sensory input, is solely constitutive of the possibility of perceiving objects as determinate spaces. (p. 22)

Since Schulting assumes, as is surely textually correct, that animals lack apperception, his view entails that animals are unable to perceive “objects as determinate spaces” (p. 22). Indeed, on his account it must be denied that animals “can have relation to an individual object, even if only indeterminately” (p. 21). Elsewhere he puts the same point in terms of complexity:

The important point here is that intuition delivers a manifold of representations (cf. B160n) but nothing beyond that in the sense of enabling a genuine perception or cognition of objects, or even a recognition of the manifold as manifold, as a plurality of representations in time, that is, as qualitatively or quantitatively complex. (p. 266; original emphasis)

But where does this leave us with animals? Recall the eagle who picks out the rapidly moving body of a mouse, who tracks that body as it passes bushes and grasses and strikes precisely it and not the similar coloured rock next to it—surely the eagle is aiming at a ‘determinate space’ and represents it as standing within a manifold of spatio-temporal and qualitatively complex relations. As Schulting himself admits at one point, it is hard to see what his alternate story of indeterminate and uncomplex spaces would really amount to:

It is difficult to understand though what it could mean […][2] to have “merely [a] manifold” (bloß Mannigfaltiges) (B160n.) of representations. It would be less than perception of determinate spatial objects, since, given the absence of any synthesis of the aggregate parts of the manifold representations, the categories enabling the determination of such objects would be lacking. (p. 326)

This certainly is hard to understand. Surely if animals can perceive spatial relations—this food is in the bowl, this food is not in the bowl—they can perceive a manifold as a manifold in some good sense of that phrase? Of course, this is not to say that they can conceptualise it as such. Indeed, the nonconceptualist’s whole point is that we need to look for new ways of cashing the ‘as’ idiom that allow for spatial determinacy understood in nonconceptual terms, rather than always falling back on an rigid link between such determinacy and conceptual synthesis.

The conclusion Schulting ultimately reaches is this:

I believe, for philosophical as well as textual reasons, that we must distinguish between representing in space, being able to orient oneself in a particular direction, as a result of e.g. a sudden noise coming from the left, or noticing some indistinct object in the distance (cf. Log, 9:33), and the representation of space or spatial objects, which latter is tantamount to the representing of something spatial as complex, namely, as located adjacent to something else and at some specified distance outside of me. Orienting oneself towards a sound coming from a particular direction requires spatial location, representing something that is still indeterminate in space, but the very act of directing one’s head left or right upon hearing a noise does not yet require the capacity to determine that indeterminate something as object, as “a determinate space” (B138), that is, as locatable in a specific place at a specific distance. (p. 309)

I think Schulting is right that this is the best end point for his argument. The claim is that animals have indistinct perception of objects—e.g. some indistinct object in the distance—but are unable to see things locatable in a specific place at a specific distance (e.g. “as located adjacent to something else”). But the danger is it now seems Burge was right: this is surely empirically false since countless sight hunters, from eagles to dogs, locate and track specific objects at specific distances with far greater acuity than the human eye. Of course, one could reply that such creatures really only have the indistinct perception Schulting credits them with and that some other mechanism explains their remarkable success. But why, given the weight of scientific evidence on the similarities between our perceptual system and theirs, should we believe this or tie Kant to it?

I suggested at the start of this section that the root of Schulting’s problem was an insistence on a binary picture of objectivity modelled on the Gegenstand/Objekt distinction. We can now see how this is borne out: as he sees it the options are either recognition of the qualitative and quantitative complexity of the manifold, where recognition is aligned with understanding, or no awareness of the manifold as a manifold at all. This makes it difficult to make sense of animals who are defined precisely by an awareness of spatio-temporal relations that falls in between those two extremes: on the one hand, they clearly can’t think of an object as ‘10ft from me’ since they lack all those concepts; on the other hand, they often can distinguish, far better than us, the difference an object being 9ft from them and 9.1ft.

Of course, Schulting is absolutely correct that the position he defends has a great deal of Kantian textual backing: that is why Allais identified it as the “dominant” view in her original article. Consider texts such as B130 with its apparent equation of all combination with that of the understanding. But matters are not as simple as that remark suggests; there is clearly some sense of ‘combination’ recognised by Kant which animals may enjoy, even if it is only associative. The nonconceptualist aim is to identify further modes of nonconceptual combination, particularly spatial combination, tacitly present in Kant’s system and which together offer a viable story regarding animal experience. I want now to close by sketching how that might be done, by stepping away from the binary choice between Gegenstand/Objekt.

3. Refining the Basic Diagnosis: Towards Multiple Notions of Objectivity

The purpose of this section is to indicate briefly how one might develop a richer model of objectivity within the First Critique. I stated above that Kant moves between multiple, non-coextensive notions of objectivity. For example, B160 talks of “space represented as object (as is really required in geometry)”. What is at stake here is a complex abstractive capacity undoubtedly beyond animals. But such a capacity is also clearly more sophisticated than mere intentionality; a being might surely have ‘object-directed’ states with respect to the material world, and lack the ability to consider space itself. What is crucial, however, is not simply to identify the various senses of objectivity in play but to track their different roles in Kant’s argument. So, for example, the Second Analogy attempts to move from this notion of objectivity:

If […] all sequence of perception would be determined solely in apprehension, i.e., merely subjectively, […] it would not thereby be objectively determined which of the perceptions must really be the preceding one and which the succeeding one. In this way, we would have only a play of representations that would not be related to any object at all. (A194/B239)

To this one:

If we investigate what new characteristic is given to our representations by the relation to an object, and what is the dignity that they thereby receive, we find that it does nothing beyond making the combination of representations necessary in a certain way, and subjecting them to a rule. (A197/B242)

The first model is defined in spatio-temporal terms, the second in terms of rules: Kant’s whole claim is that the latter makes possible the former.

Is there a master definition of objecthood operative in Kant’s system? The answer is complex and depends in part on how on one reads his claim that thought alone would be “without any object” (B146). But a good candidate for the master definition would surely be this:

Now one can to be sure call everything, and every representation, insofar as we are conscious of it, an object [Object]. But it is a question for deeper enquiry what the word “object” ought to signify with respect to appearances when these are viewed not in so far as they are (as representations) objects [Objecte], but only insofar as they stand for an object [Object]. (A189-90/B234–5)

The reference to its ‘standing for’ something meshes neatly with the modern focus on intentionality, on what Schulting called ‘object directedness’. But this alone doesn’t get us very far: as noted, there are countless theories of intentionality. For present purposes, though, we can make some progress with this simple gloss on objectivity:

A visual experience E is objective* iff E represents a distinction between spatio-temporal particulars and the mental states of the subject of that experience.

Objectivity* is, for example, key to Strawson’s reconstruction of the First Critique which defines objective experience as “experience of objects that are distinct from the experience of them” (1966:24). The question thus becomes what are the necessary and sufficient conditions for an experience to be objective*?

As I see it, what Schulting does effectively is to analyse objectivity* in terms of the notion of objectivity which Kant ties to judgment (p. 122). One characteristic feature of this approach is a tendency to over-align the categories and the logical forms of judgment. For example, he writes:

[W]hen I make a judgement about some object o, the categories must apply to o, even if I am factually wrong about the empirical content of my judgement. This has to do with the fact that categories establish the necessary unity (OUA) that is constitutive of a judgement p, regardless of the question whether p is empirically true or false. (p. 153)

Whilst a full discussion would require disambiguating and defining all the distinct notions of ‘category’ in play, I follow Longuenesse in emphasising that there are plenty of judgements which do not directly involve the categories:

[Kant] did not argue that every judgment involves the “application” of the categories that correspond to the various aspects of its logical form. Indeed, he expressly maintained the opposite view […]. For instance, the example of a hypothetical judgment Kant gives in the course of his explanation of the table of logical forms is not a causal judgment: “If there is a justice, the obstinately wicked are punished.” This judgment is a hypothetical combination of propositions founded on the analysis of the concept of ‘perfect justice’. Relying on analysis of concepts alone, without synthesis of intuition, it does not express a causal connection. (1998:78–9)

But the real problem is not this, partly since this type of issue can always be managed by deploying multiple senses of ‘category’. Rather the danger is a tendency to miss the fact that, whilst the forms of objectivity Kant links to marks, rules and judgements are sufficient for objective*, they are not necessary.

This is because there are other ‘modes of givenness’, to borrow a phenomenological phrasing, that allow the objective* distinction to be experientially manifest without its being conceptually marked.[3] Smith provides a neat formulation of one version of the view:

Perception concerns the ‘external world’. The suggestion is that this is, in essential part, because perceptual experience presents ‘external’ objects as literally external—to our bodies. A bodily sensation such as a headache is experienced as in your head; it is not perceived as an object with your head. When, by contrast, you look at your hand, although the object seen is not spatially separated from you (since it is a part of you), it is, nevertheless, spatially separate from the eye with which (and from where) you see it. (2002:134)

The claim is that a three dimensional egocentrically orientated awareness of space within which something is seen as more or less distant is sufficient to sustain a distinction between spatio-temporal particulars and the subject’s own states, such as sensations.

Of course, animals do not have concepts such as <external world> or <sensation>, nor do they have the type of cognitive architecture associated with judgement’s ability to posit connections regardless of “any difference in the condition of the subject” (B141–2). Rather, the proposal is that the way in which entities are given to them as spatially arrayed is sufficiently distinct from the way in which sensations are given to them that it constitutes a distinctly perceptual or intuitive, as opposed to sensory, mode of experience, one which is well described as ‘object-directed’. So, for example, insofar as the animal encounters something as arrayed within such a space, it is given only from a single perspective, a perspective which changes as the object gradually unfolds in line with the animal’s movements and motor dispositions. In contrast, sensation is non-perspectival—whilst the dog experiences the pain in its foot, as opposed to its leg, there is no angle from which it does so. In short, there is a distinctive and phenomenologically articulable mode of givenness that allows one to legitimately ascribe the object* distinction to animals, even though, of course, they cannot articulate it.

If we now return to nonconceptualism, one can see how a genuine compromise on the issue is possible. On the one hand, animals have spatio-temporal awareness of particulars at whatever level of resolution or determinacy the biology suggests. In virtue of the relevant mode of givenness, this awareness is objective*. Animal intentionality is thus more determinate, structured and sophisticated than Schulting allows. On the other hand, however, there are clearly other forms of objective awareness that animals lack, such as those associated with judgement. Recall the example I gave from the Second Analogy at the start of this section. As I see it, Kant’s claim in the Deduction and Principles is not that the ability to perceive determinate spatio-temporal particulars depends on judgement, but that the ability to perceive a quite specific subset of spatio-temporal relations does. One such relation is the relation of objective succession. A sighthound can readily perceive multiple particulars in a high level of resolution and as standing in complex spatial relations. But what it is unable to track is the specific distinction between perception of succession and succession of perception.

In short, Kant’s project works by interweaving a wide range of notions of objectivity, some functioning as premises, some as conclusions. This interweaving is central to arguments such as the Deduction—and the significance of nonconceptualism is in part that it highlights that fact. Ultimately, without the kind of move defended above, the vast, elaborate, baroque, genius of the First Critique would be vulnerable to as mundane a counterexample as a YouTube dog.

Invited: 2 April 2017; received: 1 December 2017.


[1] It is not a claim, I, as a nonconceptualist, have ever defended.

[2] I have excised a remark on synthesis to avoid bringing in the additional exegetical claim made by Schulting that all synthesis is conceptual; I touch on this at the end of the current section.

[3] I develop the following line of argument in detail in Golob (forthcoming).


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

Burge, T. (2010), The Origins of Objectivity (Oxford: Oxford University Press).

Falkenstein, L. (2006), ‘Kant’s Transcendental Aesthetic’, in G. Bird (ed.), A Companion to Kant (Oxford: Blackwell), pp. 140–53.

Ginsborg, H. (2008), ‘Was Kant a Nonconceptualist?’, Philosophical Studies 137: 65–77.

Golob, S. (forthcoming), ‘What do animals see?’, in L. Allais & J. Callanan (eds), Kant and Animals (Oxford: Oxford University Press).

Hanna, R. (2008), ‘Kantian Non-Conceptualism’, Philosophical Studies 137: 41–64.

Hume, D. (1978), A Treatise of Human Nature, ed. P. Nidditch (Oxford: Clarendon Press).

Longuenesse, B. (1998), Kant and the Capacity to Judge (Princeton, NJ: Princeton University Press).

Smith, A.D. (2002), The Problem of Perception (Cambridge, MA: Harvard UniversityPress).

Strawson, P.F. (1966), The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason (London: Routledge).

© Sacha Golob, 2018.

Sacha Golob is a Senior Lecturer in Philosophy at King’s College, London. His research focuses on the intersection between Kantian and post-Kantian philosophy and contemporary work on the philosophy of mind, aesthetics and philosophical methodology. Among publications in British Journal of the History of PhilosophyEuropean Journal of PhilosophyKantian ReviewProceedings of the Aristotelian Society and in various edited volumes, he is the author of Heidegger on Concepts, Freedom and Normativity (Cambridge UP, 2014) and co-editor of the Cambridge History of Moral Philosophy (Cambridge UP, 2017).