In Kant’s Radical Subjectivism, Dennis Schulting considers aspects of the transcendental deduction of the categories not discussed in his (Reference Schulting2013). He also amplifies points from that earlier volume – for example, his claim that the deduction derives each individual category directly from apperception, and his view that the deduction argues both progressively and regressively. He stresses that Kant is a ‘radical subjectivist’ who holds that ‘the very constitution of the object as object’ depends on apperception (pp. 10, 11).Footnote
1
For Schulting, the deduction thus provides reasons independent of the Transcendental Aesthetic for accepting transcendental idealism. He ends with the radical view that Kant’s view of discursive concepts itself requires idealism. Along the way, he considers many other topics.
Earlier reviewers have, I think rightly, criticized Schulting’s attempt to derive the individual categories from apperception.Footnote
2
But his view of the progressive-regressive structure of the Deduction also raises important questions, and I consider it below. I then turn to his claims about object constitution and finally to the idea that discursivity implies idealism.
1. The Deduction: Progressive? Regressive?
The deduction’s main goal is to explain and justify the objective validity of the categories, the subsumption of all objects that we can cognize under those a priori concepts (A84–5/B116–17). In the Prolegomena, Kant says that the first Critique – and so presumably the deduction, that book’s central reasoning – argues progressively rather than regressively.Footnote
3
Progressive reasoning in general argues for a claim by deducing it from premises accepted by everyone who may question that claim. Regressive reasoning assumes the claim and identifies premises from which that claim can in some deductive or other way be shown to follow (for example, as an explanatory consequence of the premises). As indicated below, the Prolegomena amplifies these descriptions, but they provide a good starting point.
Standard progressive interpretations see the deduction as beginning with the necessary claim that unity of apperception holds with respect to the manifold of any intuition through which we cognize an object. The object cognition introduced here must allow for the possibility (which the argument will subsequently eliminate) that the object cognized is not category-subsumed. From the holding of unity of apperception, Kant infers the synthesis of the manifold and the category-subsumption of the object known through that manifold. This object is a category-structured and therefore robust entity (robust in the sense of being distinct from the elements of the manifold themselves or any merely associatively ordered arrangement of those elements in our mind). Every object that we cognize through a sensible intuition (and we cognize all objects through such intuitions) is therefore category-subsumed.
Regressive interpretations take the deduction to begin by assuming that, through the manifold, we cognize a thing that from the start we regard as a robust object (for example, an ordinary spatiotemporal thing). The deduction then argues that the nature of that cognition requires the object to be category-subsumed, and it explains how, given Kant’s cognitive framework, that category-subsumption occurs (cf. Ameriks Reference Ameriks1978).
Progressive interpretations usually take Kant to be refuting scepticism about the validity of the categories with respect to every object we can cognize – or even to be refuting scepticism about our knowledge of the external world. Regressive interpretations take Kant not to target scepticism. Rather, he simply assumes we know robust objects and develops the commitment to category-subsumption that that assumption carries.
Schulting holds that the deduction is both regressive and progressive. The main argument is regressive (pp. 73–4, 77–8; Schulting Reference Schulting2013: 61ff., 212). Kant assumes we cognize robust spatiotemporal objects. He shows that such cognition necessitates an objective unity in the manifold, a unity that requires the object cognized to be category-subsumed. Kant’s ‘modest aim’ (p. 213) is not to refute sceptics but to analyse such cognition and show how it is possible (pp. 58, 301, 297; Schulting Reference Schulting2013: 211).
This regressive argument does not, however, give philosophical legitimacy to the conclusion that the (robust) objects here cognized are category-subsumed. This conclusion must be proved ‘valid according to a rationally insightful method, which lends it apodicticity … and universal generality’ (Schulting Reference Schulting2013: 212; cf. Reference Schulting2017: 57–8). Progressive reasoning gives that proof, beginning with the holding of unity of apperception. Kant then deductively infers that the elements of the manifold form an objective unity in such a way that the (robust) object cognized through that manifold is category-subsumed.
Schulting here provides an attractive way for regressive interpretations to acknowledge the deduction’s reasoning from apperception to category-subsumption. However, I think he misreads the deduction’s structure. As both the text and the deduction’s argumentative goals show, the deduction is fundamentally progressive. Its regressive sections simply provide preliminary reasoning that introduces Kant’s cognitive apparatus (the manifold, synthesis, imagination, apperception, the categories), describes how that apparatus operates and shows how that operation requires category application. But it is the progressive argument that achieves (and is meant by Kant to achieve) the main goal of the deduction, the a priori demonstration that, necessarily, any object whatsoever that we may cognize – and not just any robust object that we in fact cognize – is category-subsumed.
The basic contours of this interpretation are supported from many directions. One need not regard Kant as obsessed with external-world scepticism to see that in the deduction he wants to show the necessity that all objects that we may cognize fall under the categories.Footnote
4
Of course this point includes all robust objects. But the deduction also must rule out the possibility that we might cognize objects only of a very minimal sort (for example, the merely associatively organized sequences of B127), objects that simply are not category-subsumed in our cognition of them. After all, Hume allows this possibility; and texts such as B127–8 show that, at the beginning of the deduction, Kant is alive to what he calls Hume’s ‘scepticism’ that a priori concepts such as the category of causality have any objective application to objects, rather than a mere origin in custom and association.Footnote
5
Many other texts in the Transcendental Deduction support this point. Thus note B164–5, where Kant stresses that he proves that ‘all possible perceptions, hence everything that can ever reach empirical consciousness … stand under the categories’ (my emphases), or A107. Moreover, Schulting’s own identifications of regressive and progressive parts of the deduction do not strongly support his position. As regressive texts, Schulting explicitly cites only preliminary passages such as A92–4/B124–6 and the threefold synthesis.Footnote
6
The former is an introductory text, and Kant here just assumes without argument that the application in experience of the concept of an object involves category application. The threefold synthesis argues that cognition of an object requires the manifold of intuition to be synthesized via apprehension, reproduction and recognition. Then Kant argues that the object cognized is category-subsumed. This reasoning is, overall, regressive. But the threefold synthesis is still an introductory investigation, in which Kant wants simply ‘to prepare [rather than] to instruct the reader’ (A98).
On the progressive side, Schulting appears to regard the entire B-Deduction reasoning, as far as it argues for the objective validity of the categories, as progressive.Footnote
7
So, for Schulting, none of the actual official B-Deduction seems to be centrally regressive. But Schulting takes the main deduction argument to be regressive. Hence he appears committed to holding that the B-Deduction, Kant’s carefully considered replacement for the A-Deduction, does not contain Kant’s main argument for the objective validity of the categories at all. This result is most implausible. Is this really Schulting’s view?
Finally, Prolegomena §4 stresses that the first Critique argument proceeds synthetically (progressively) by developing knowledge of its conclusions ‘out of [pure reason’s] original seeds without relying on any fact whatever’ (4: 274). In regressive texts we ‘ascend’ from ‘something already known’ to ‘the sources … whose discovery … will explain what is known already’ (4: 275). But starting from a fact, something already known that we want to explain (e.g. cognition of a robust object), is what regressive reasoning does, not the progressive deduction reasoning that starts with the necessary, a priori holding of unity of apperception and develops a proof of the categories’ objective validity in an a priori manner.
I have not discussed all the ins and outs of regressive readings, and more can be said in their defence. But I remain convinced that, as I argued in KTD – and as many others argue – the central deduction argument is progressive.
2. Objectivity, Gaps
Schulting’s claims for Kant’s ‘radical subjectivism’, the ‘constitution of the object as object’ by apperception, are part of his larger discussion of synthesis and the categories. This discussion includes his valuable chapter 7 account of B-Deduction §24 on figurative synthesis. Here, however, I focus on object constitution.
Schulting’s basic position is that the object, insofar as we cognize it,Footnote
8
is category-subsumed; and its being so subsumed is a property that it has only ‘due to the subject’s [our own] agency’ (p. 14). Thus the object’s formal, structural property of being a substance and so of having a subject-predicate structure is present in the object ultimately only because of the holding of unity of apperception in the synthesis of the manifold in cognition. Idealism holds of the cognized object, for that object simply is the thing in the content of our cognition that has that formal property, and it has that property only through apperceptive synthesis. So, Schulting concludes, contrary to interpreters such as Allais, the deduction is not purely epistemic. It has transcendental-idealist consequences.
However one develops the above view, Schulting is right that Kant himself would endorse the overall position that I have just sketched, with its included idealism. But whether this position, in Kant’s hands, is the product of an argument for idealism entirely independent of the Aesthetic, whether Kant himself gives that argument and whether that argument is, as Schulting holds, gap-free, are questions needing further discussion.
Schulting’s reasoning appears to run thus (pp. 167–9, my exposition). Suppose I cognize an object via a manifold of representations r, s, t … Because unity of apperception holds with respect to that manifold, the synthesized group r, s, t …, with the tag ‘mine’, occurs before my mind. So occurring, this group functions for me as an object (of thought). So I apply to it, veridically, the Kantian concept of an object in general. But the categories provide rules for the synthesis that yields this synthesized group. So this group, this object before my consciousness, is category-subsumed. I have no direct cognitive access to anything beyond my representations, however, so whatever thing it is that I cognize through the r, s, t … manifold, that thing cannot exist in itself (p. 169). Indeed, that thing can only be this category-subsumed group itself, which is indeed an object of which I am veridically conscious. So the object I cognize falls under the categories, and that object has its categorial structure simply because of my apperceptive synthesis. A Schulting-style object-constitution thus holds, and the object cognized is cognition-dependent. Nor are there any of the familiar gaps in this reasoning between, say, how I conceive of this object and how it actually is.
Kant perhaps gives something like this reasoning at A104–10 of the threefold synthesis, where he asserts that we have nothing that exists, in itself, outside our representations that we could take to be the object cognized. Then (to skip details) he argues that those representations must be subject to unity of apperception in such a way that through them we cognize a category-subsumed object. Schulting’s reasoning is an idealized reconstruction that need not capture every detail of any particular text, and it suggests an interesting reading of A104–10.
But is this reasoning gap-free, and does it provide an Aesthetic-independent argument that Kant himself gives for idealism? Here I have doubts. To note two major concerns:
-
(i) Kant certainly claims that the categories provide rules for synthesis.Footnote
9
But why must the synthesis that unites r, s, t … before my mind with the tag ‘mine’ be, specifically, category-governed? Why may not the object that I cognize via those representations simply be those representations themselves, so tagged? As far as I can see, neither Kant nor Schulting answers such questions convincingly.
-
(ii) Schulting holds that because I have no direct cognitive access to anything beyond my representations, the object I cognize cannot exist in itself beyond those representations. But this reasoning succeeds only if my representations yield no knowledge of objects in themselves. To show that point, Schulting cannot now appeal to the Aesthetic arguments without making the deduction’s idealist conclusion depend on the Aesthetic’s. Moreover, at A104 Kant seems clearly to refer to Aesthetic results before making his no-object-outside-our-representations assertion.Footnote
10
So I doubt that Kant, at least here, follows the Aesthetic-independent reading of this part of Schulting’s argument. Finally, if we drop the no-object-outside-our-representations reasoning, then Schulting’s argument allows the object cognized to exist in itself. But then why must the categorial structure that we take to belong to the synthesized intuition belong also to the object that we know through that intuition?
Schulting’s argument thus has gaps and appears not to offer a piece of Aesthetic-independent reasoning that Kant himself defends. Nevertheless, I think Schulting is right in general to stress what he calls ‘Kant’s subjectivism’. The deduction reasoning from unity of apperception, when freely reconstructed with plausible premises and valid inferences, may not require idealism, however successful it may otherwise be. But Kant himself certainly understands the deduction to involve idealism. He sees the cognitive subject’s apperception as the ground of category structure in the object cognized, and interpreters should not soft-pedal this fact.
Finally, Schulting also is absolutely right that Kant regards the deduction as providing an Aesthetic-independent argument for idealism, although Schulting’s discussion seems to bypass the clearest evidence for this claim. The key text, which I do not recall Schulting considering in connection with this topic, is B167–8, where Kant holds that the kind of necessary unity in the object required by the category-subsumption that the deduction proves would be uncognizable by us if it belonged to an object existing in itself.Footnote
11
3. Discursivity and Idealism
At the end of his book Schulting pushes his radical-subjectivist interpretation even further. He argues that Kant’s account of discursive, general concepts by itself requires a kind of idealism. Schulting’s argument focuses on Kant’s concept of an object. Schulting holds that, using that concept, one cannot form true thoughts (let alone cognition-embodying thoughts) about a genuine realm of determinate objects existing in themselves. We do succeed in thinking using that concept. But whatever it is that we then think, the nature of our discursivity means that we have no access to the nature of a thing in itself. The idealism here concerns not cognition-dependency but ‘the inherent limitations of discursive thinking … as a result of which we cannot have a determinate concept of a thing in itself’ (p. 374).
Schulting appears to reason thus. For Kant, our discursive, general concepts simply present us with finite lists of general properties (e.g. of being iron and being cubical). But conceptually, a thing in itself is fully determinate (for each property in the list of possible properties, that thing has that property or that thing has the negation of that property). For us to think about a thing in itself ‘as it is in itself’ (p. 393), ‘as a thoroughly determined individual’ (p. 404), we would have to specify, in our concept of a thing in itself, the infinite list of properties that picks that thing out. But our human concepts present only finite lists of properties, so we cannot do this. Whatever we grasp with our concept of an object in itself is therefore not ‘the thing in itself as such’ (p. 381). We have no definite notion of what a thing in itself might be (p. 383; cf. 374).
I think that Schulting’s reasoning here is mistaken.Footnote
12
For Kant, concepts are indeed general representations, as described above. Through concepts taken by themselves we can think only general thoughts that apply to the various objects that have the properties presented by those concepts (so-called de dicto thoughts). Using only the concepts of being an aardvark and being nocturnal, we can think (judge) that all or some aardvarks are nocturnal, for example. But we do so without thinking, of any particular aardvark, say Jolene in the Cincinnati zoo, that she is nocturnal. What, for Kant, allows us in thoughts to refer to and predicate properties of particular, individual things such as Jolene (so-called de re thoughts) is our intuitions. Intuitions represent single, particular, individuated objects as such. Combining a Jolene-intuition with our concepts, we can think, of Jolene, that she is nocturnal, and so on.
There is a catch, however. Given transcendental idealism, beings like us have only sensible intuitions, and these intuitions represent particular objects only as they appear to us spatiotemporally, not as they are in themselves. We can use an intuition of Jolene to pick her out as a particular aardvarkian appearance and then apply to her our concepts so as to judge, of her, that she is a nocturnal aardvark. But we have no intuitions that represent to us particular objects as they are in themselves. Hence we cannot judge, of any particular thing in itself, that that thing is such-and-such. Here Schulting is right: our concept of a thing in itself does not allow us to grasp any particular things in themselves as the particular entities that they are.
However – and here Schulting goes wrong – nothing in Kant’s views here prevents us from using general concepts by themselves to make true judgements about the various particular objects that fall under those concepts. Using just our aardvark and nocturnal concepts, we can judge truly that all aardvarks are nocturnal. For this judgement to be true, all that is required is that every object that has the aardvark property (and so falls under that concept) also has the nocturnal property (and so falls under that concept). No referential relation between us (or between our concepts here) and any particular aardvarks is required. The same is true for our use of the general concept of a thing in itself. Using that concept, we can make what for Kant is the true general judgement that every thing in itself is non-spatiotemporal. The objects that this judgement concerns are the fully determinate, cognition-independent entities that Kant calls things in themselves. Our judgement is true because (given Kant’s views) any object that has the property of being such an entity also has the property of being non-spatiotemporal. This judgement is true even though we have, and can have, no referential relations to any individual such objects that pick them out as the particular, fully determined, infinitely propertied things that, in themselves, they are.Footnote
13
As far as I can see, then, Schulting’s discussion has taken a wrong turn. A general, de dicto judgement such as ‘no thing in itself is spatial’ is about the very things in themselves that Kant takes to ground our knowledge. That judgement concerns those things even though it picks out no particular such things ‘as such’.Footnote
14
Discursivity simply does not imply idealism.
Schulting’s work is sometimes difficult to follow. But he raises interesting questions. As noted, I think his progressive-regressive views are mistaken; his discussion of apperception and objectivity leaves argumentative gaps; and discursivity does not imply idealism. However, he is right that Kant accepts object-constitution in an idealist way. He is also right in his general claim that in the deduction Kant provides an argument for transcendental idealism that is independent of the Aesthetic’s arguments. Along the way, he makes valuable points about specific Kantian topics, e.g. B-Deduction §24’s figurative synthesis.
In Kant’s Radical Subjectivism, Dennis Schulting considers aspects of the transcendental deduction of the categories not discussed in his (Reference Schulting2013). He also amplifies points from that earlier volume – for example, his claim that the deduction derives each individual category directly from apperception, and his view that the deduction argues both progressively and regressively. He stresses that Kant is a ‘radical subjectivist’ who holds that ‘the very constitution of the object as object’ depends on apperception (pp. 10, 11).Footnote 1 For Schulting, the deduction thus provides reasons independent of the Transcendental Aesthetic for accepting transcendental idealism. He ends with the radical view that Kant’s view of discursive concepts itself requires idealism. Along the way, he considers many other topics.
Earlier reviewers have, I think rightly, criticized Schulting’s attempt to derive the individual categories from apperception.Footnote 2 But his view of the progressive-regressive structure of the Deduction also raises important questions, and I consider it below. I then turn to his claims about object constitution and finally to the idea that discursivity implies idealism.
1. The Deduction: Progressive? Regressive?
The deduction’s main goal is to explain and justify the objective validity of the categories, the subsumption of all objects that we can cognize under those a priori concepts (A84–5/B116–17). In the Prolegomena, Kant says that the first Critique – and so presumably the deduction, that book’s central reasoning – argues progressively rather than regressively.Footnote 3 Progressive reasoning in general argues for a claim by deducing it from premises accepted by everyone who may question that claim. Regressive reasoning assumes the claim and identifies premises from which that claim can in some deductive or other way be shown to follow (for example, as an explanatory consequence of the premises). As indicated below, the Prolegomena amplifies these descriptions, but they provide a good starting point.
Standard progressive interpretations see the deduction as beginning with the necessary claim that unity of apperception holds with respect to the manifold of any intuition through which we cognize an object. The object cognition introduced here must allow for the possibility (which the argument will subsequently eliminate) that the object cognized is not category-subsumed. From the holding of unity of apperception, Kant infers the synthesis of the manifold and the category-subsumption of the object known through that manifold. This object is a category-structured and therefore robust entity (robust in the sense of being distinct from the elements of the manifold themselves or any merely associatively ordered arrangement of those elements in our mind). Every object that we cognize through a sensible intuition (and we cognize all objects through such intuitions) is therefore category-subsumed.
Regressive interpretations take the deduction to begin by assuming that, through the manifold, we cognize a thing that from the start we regard as a robust object (for example, an ordinary spatiotemporal thing). The deduction then argues that the nature of that cognition requires the object to be category-subsumed, and it explains how, given Kant’s cognitive framework, that category-subsumption occurs (cf. Ameriks Reference Ameriks1978).
Progressive interpretations usually take Kant to be refuting scepticism about the validity of the categories with respect to every object we can cognize – or even to be refuting scepticism about our knowledge of the external world. Regressive interpretations take Kant not to target scepticism. Rather, he simply assumes we know robust objects and develops the commitment to category-subsumption that that assumption carries.
Schulting holds that the deduction is both regressive and progressive. The main argument is regressive (pp. 73–4, 77–8; Schulting Reference Schulting2013: 61ff., 212). Kant assumes we cognize robust spatiotemporal objects. He shows that such cognition necessitates an objective unity in the manifold, a unity that requires the object cognized to be category-subsumed. Kant’s ‘modest aim’ (p. 213) is not to refute sceptics but to analyse such cognition and show how it is possible (pp. 58, 301, 297; Schulting Reference Schulting2013: 211).
This regressive argument does not, however, give philosophical legitimacy to the conclusion that the (robust) objects here cognized are category-subsumed. This conclusion must be proved ‘valid according to a rationally insightful method, which lends it apodicticity … and universal generality’ (Schulting Reference Schulting2013: 212; cf. Reference Schulting2017: 57–8). Progressive reasoning gives that proof, beginning with the holding of unity of apperception. Kant then deductively infers that the elements of the manifold form an objective unity in such a way that the (robust) object cognized through that manifold is category-subsumed.
Schulting here provides an attractive way for regressive interpretations to acknowledge the deduction’s reasoning from apperception to category-subsumption. However, I think he misreads the deduction’s structure. As both the text and the deduction’s argumentative goals show, the deduction is fundamentally progressive. Its regressive sections simply provide preliminary reasoning that introduces Kant’s cognitive apparatus (the manifold, synthesis, imagination, apperception, the categories), describes how that apparatus operates and shows how that operation requires category application. But it is the progressive argument that achieves (and is meant by Kant to achieve) the main goal of the deduction, the a priori demonstration that, necessarily, any object whatsoever that we may cognize – and not just any robust object that we in fact cognize – is category-subsumed.
The basic contours of this interpretation are supported from many directions. One need not regard Kant as obsessed with external-world scepticism to see that in the deduction he wants to show the necessity that all objects that we may cognize fall under the categories.Footnote 4 Of course this point includes all robust objects. But the deduction also must rule out the possibility that we might cognize objects only of a very minimal sort (for example, the merely associatively organized sequences of B127), objects that simply are not category-subsumed in our cognition of them. After all, Hume allows this possibility; and texts such as B127–8 show that, at the beginning of the deduction, Kant is alive to what he calls Hume’s ‘scepticism’ that a priori concepts such as the category of causality have any objective application to objects, rather than a mere origin in custom and association.Footnote 5
Many other texts in the Transcendental Deduction support this point. Thus note B164–5, where Kant stresses that he proves that ‘all possible perceptions, hence everything that can ever reach empirical consciousness … stand under the categories’ (my emphases), or A107. Moreover, Schulting’s own identifications of regressive and progressive parts of the deduction do not strongly support his position. As regressive texts, Schulting explicitly cites only preliminary passages such as A92–4/B124–6 and the threefold synthesis.Footnote 6 The former is an introductory text, and Kant here just assumes without argument that the application in experience of the concept of an object involves category application. The threefold synthesis argues that cognition of an object requires the manifold of intuition to be synthesized via apprehension, reproduction and recognition. Then Kant argues that the object cognized is category-subsumed. This reasoning is, overall, regressive. But the threefold synthesis is still an introductory investigation, in which Kant wants simply ‘to prepare [rather than] to instruct the reader’ (A98).
On the progressive side, Schulting appears to regard the entire B-Deduction reasoning, as far as it argues for the objective validity of the categories, as progressive.Footnote 7 So, for Schulting, none of the actual official B-Deduction seems to be centrally regressive. But Schulting takes the main deduction argument to be regressive. Hence he appears committed to holding that the B-Deduction, Kant’s carefully considered replacement for the A-Deduction, does not contain Kant’s main argument for the objective validity of the categories at all. This result is most implausible. Is this really Schulting’s view?
Finally, Prolegomena §4 stresses that the first Critique argument proceeds synthetically (progressively) by developing knowledge of its conclusions ‘out of [pure reason’s] original seeds without relying on any fact whatever’ (4: 274). In regressive texts we ‘ascend’ from ‘something already known’ to ‘the sources … whose discovery … will explain what is known already’ (4: 275). But starting from a fact, something already known that we want to explain (e.g. cognition of a robust object), is what regressive reasoning does, not the progressive deduction reasoning that starts with the necessary, a priori holding of unity of apperception and develops a proof of the categories’ objective validity in an a priori manner.
I have not discussed all the ins and outs of regressive readings, and more can be said in their defence. But I remain convinced that, as I argued in KTD – and as many others argue – the central deduction argument is progressive.
2. Objectivity, Gaps
Schulting’s claims for Kant’s ‘radical subjectivism’, the ‘constitution of the object as object’ by apperception, are part of his larger discussion of synthesis and the categories. This discussion includes his valuable chapter 7 account of B-Deduction §24 on figurative synthesis. Here, however, I focus on object constitution.
Schulting’s basic position is that the object, insofar as we cognize it,Footnote 8 is category-subsumed; and its being so subsumed is a property that it has only ‘due to the subject’s [our own] agency’ (p. 14). Thus the object’s formal, structural property of being a substance and so of having a subject-predicate structure is present in the object ultimately only because of the holding of unity of apperception in the synthesis of the manifold in cognition. Idealism holds of the cognized object, for that object simply is the thing in the content of our cognition that has that formal property, and it has that property only through apperceptive synthesis. So, Schulting concludes, contrary to interpreters such as Allais, the deduction is not purely epistemic. It has transcendental-idealist consequences.
However one develops the above view, Schulting is right that Kant himself would endorse the overall position that I have just sketched, with its included idealism. But whether this position, in Kant’s hands, is the product of an argument for idealism entirely independent of the Aesthetic, whether Kant himself gives that argument and whether that argument is, as Schulting holds, gap-free, are questions needing further discussion.
Schulting’s reasoning appears to run thus (pp. 167–9, my exposition). Suppose I cognize an object via a manifold of representations r, s, t … Because unity of apperception holds with respect to that manifold, the synthesized group r, s, t …, with the tag ‘mine’, occurs before my mind. So occurring, this group functions for me as an object (of thought). So I apply to it, veridically, the Kantian concept of an object in general. But the categories provide rules for the synthesis that yields this synthesized group. So this group, this object before my consciousness, is category-subsumed. I have no direct cognitive access to anything beyond my representations, however, so whatever thing it is that I cognize through the r, s, t … manifold, that thing cannot exist in itself (p. 169). Indeed, that thing can only be this category-subsumed group itself, which is indeed an object of which I am veridically conscious. So the object I cognize falls under the categories, and that object has its categorial structure simply because of my apperceptive synthesis. A Schulting-style object-constitution thus holds, and the object cognized is cognition-dependent. Nor are there any of the familiar gaps in this reasoning between, say, how I conceive of this object and how it actually is.
Kant perhaps gives something like this reasoning at A104–10 of the threefold synthesis, where he asserts that we have nothing that exists, in itself, outside our representations that we could take to be the object cognized. Then (to skip details) he argues that those representations must be subject to unity of apperception in such a way that through them we cognize a category-subsumed object. Schulting’s reasoning is an idealized reconstruction that need not capture every detail of any particular text, and it suggests an interesting reading of A104–10.
But is this reasoning gap-free, and does it provide an Aesthetic-independent argument that Kant himself gives for idealism? Here I have doubts. To note two major concerns:
(i) Kant certainly claims that the categories provide rules for synthesis.Footnote 9 But why must the synthesis that unites r, s, t … before my mind with the tag ‘mine’ be, specifically, category-governed? Why may not the object that I cognize via those representations simply be those representations themselves, so tagged? As far as I can see, neither Kant nor Schulting answers such questions convincingly.
(ii) Schulting holds that because I have no direct cognitive access to anything beyond my representations, the object I cognize cannot exist in itself beyond those representations. But this reasoning succeeds only if my representations yield no knowledge of objects in themselves. To show that point, Schulting cannot now appeal to the Aesthetic arguments without making the deduction’s idealist conclusion depend on the Aesthetic’s. Moreover, at A104 Kant seems clearly to refer to Aesthetic results before making his no-object-outside-our-representations assertion.Footnote 10 So I doubt that Kant, at least here, follows the Aesthetic-independent reading of this part of Schulting’s argument. Finally, if we drop the no-object-outside-our-representations reasoning, then Schulting’s argument allows the object cognized to exist in itself. But then why must the categorial structure that we take to belong to the synthesized intuition belong also to the object that we know through that intuition?
Schulting’s argument thus has gaps and appears not to offer a piece of Aesthetic-independent reasoning that Kant himself defends. Nevertheless, I think Schulting is right in general to stress what he calls ‘Kant’s subjectivism’. The deduction reasoning from unity of apperception, when freely reconstructed with plausible premises and valid inferences, may not require idealism, however successful it may otherwise be. But Kant himself certainly understands the deduction to involve idealism. He sees the cognitive subject’s apperception as the ground of category structure in the object cognized, and interpreters should not soft-pedal this fact.
Finally, Schulting also is absolutely right that Kant regards the deduction as providing an Aesthetic-independent argument for idealism, although Schulting’s discussion seems to bypass the clearest evidence for this claim. The key text, which I do not recall Schulting considering in connection with this topic, is B167–8, where Kant holds that the kind of necessary unity in the object required by the category-subsumption that the deduction proves would be uncognizable by us if it belonged to an object existing in itself.Footnote 11
3. Discursivity and Idealism
At the end of his book Schulting pushes his radical-subjectivist interpretation even further. He argues that Kant’s account of discursive, general concepts by itself requires a kind of idealism. Schulting’s argument focuses on Kant’s concept of an object. Schulting holds that, using that concept, one cannot form true thoughts (let alone cognition-embodying thoughts) about a genuine realm of determinate objects existing in themselves. We do succeed in thinking using that concept. But whatever it is that we then think, the nature of our discursivity means that we have no access to the nature of a thing in itself. The idealism here concerns not cognition-dependency but ‘the inherent limitations of discursive thinking … as a result of which we cannot have a determinate concept of a thing in itself’ (p. 374).
Schulting appears to reason thus. For Kant, our discursive, general concepts simply present us with finite lists of general properties (e.g. of being iron and being cubical). But conceptually, a thing in itself is fully determinate (for each property in the list of possible properties, that thing has that property or that thing has the negation of that property). For us to think about a thing in itself ‘as it is in itself’ (p. 393), ‘as a thoroughly determined individual’ (p. 404), we would have to specify, in our concept of a thing in itself, the infinite list of properties that picks that thing out. But our human concepts present only finite lists of properties, so we cannot do this. Whatever we grasp with our concept of an object in itself is therefore not ‘the thing in itself as such’ (p. 381). We have no definite notion of what a thing in itself might be (p. 383; cf. 374).
I think that Schulting’s reasoning here is mistaken.Footnote 12 For Kant, concepts are indeed general representations, as described above. Through concepts taken by themselves we can think only general thoughts that apply to the various objects that have the properties presented by those concepts (so-called de dicto thoughts). Using only the concepts of being an aardvark and being nocturnal, we can think (judge) that all or some aardvarks are nocturnal, for example. But we do so without thinking, of any particular aardvark, say Jolene in the Cincinnati zoo, that she is nocturnal. What, for Kant, allows us in thoughts to refer to and predicate properties of particular, individual things such as Jolene (so-called de re thoughts) is our intuitions. Intuitions represent single, particular, individuated objects as such. Combining a Jolene-intuition with our concepts, we can think, of Jolene, that she is nocturnal, and so on.
There is a catch, however. Given transcendental idealism, beings like us have only sensible intuitions, and these intuitions represent particular objects only as they appear to us spatiotemporally, not as they are in themselves. We can use an intuition of Jolene to pick her out as a particular aardvarkian appearance and then apply to her our concepts so as to judge, of her, that she is a nocturnal aardvark. But we have no intuitions that represent to us particular objects as they are in themselves. Hence we cannot judge, of any particular thing in itself, that that thing is such-and-such. Here Schulting is right: our concept of a thing in itself does not allow us to grasp any particular things in themselves as the particular entities that they are.
However – and here Schulting goes wrong – nothing in Kant’s views here prevents us from using general concepts by themselves to make true judgements about the various particular objects that fall under those concepts. Using just our aardvark and nocturnal concepts, we can judge truly that all aardvarks are nocturnal. For this judgement to be true, all that is required is that every object that has the aardvark property (and so falls under that concept) also has the nocturnal property (and so falls under that concept). No referential relation between us (or between our concepts here) and any particular aardvarks is required. The same is true for our use of the general concept of a thing in itself. Using that concept, we can make what for Kant is the true general judgement that every thing in itself is non-spatiotemporal. The objects that this judgement concerns are the fully determinate, cognition-independent entities that Kant calls things in themselves. Our judgement is true because (given Kant’s views) any object that has the property of being such an entity also has the property of being non-spatiotemporal. This judgement is true even though we have, and can have, no referential relations to any individual such objects that pick them out as the particular, fully determined, infinitely propertied things that, in themselves, they are.Footnote 13
As far as I can see, then, Schulting’s discussion has taken a wrong turn. A general, de dicto judgement such as ‘no thing in itself is spatial’ is about the very things in themselves that Kant takes to ground our knowledge. That judgement concerns those things even though it picks out no particular such things ‘as such’.Footnote 14 Discursivity simply does not imply idealism.
Schulting’s work is sometimes difficult to follow. But he raises interesting questions. As noted, I think his progressive-regressive views are mistaken; his discussion of apperception and objectivity leaves argumentative gaps; and discursivity does not imply idealism. However, he is right that Kant accepts object-constitution in an idealist way. He is also right in his general claim that in the deduction Kant provides an argument for transcendental idealism that is independent of the Aesthetic’s arguments. Along the way, he makes valuable points about specific Kantian topics, e.g. B-Deduction §24’s figurative synthesis.