1

According to the non-conceptualist interpretation, what Kant is doing in the transcendental deduction is taking it for granted that there are objects that we think about, and defending a claim about the necessary conditions of thinking or making a judgment about a given object. The claim is that one of the necessary conditions of thinking about a given object is the application of the categories or pure concepts of the understanding to this object. We can distinguish two versions of the non-conceptualist interpretation according to the way this necessary condition is spelled out. According to the first version of the interpretation, what Kant is arguing in the transcendental deduction is that one of the necessary conditions of thinking about a given object is the instantiation of the categories by this object. If we are thinking about a given object, this object must be a unity, or a plurality, or a totality, or a reality, and so on for each of the other categories. According to the second version of the non-conceptualist interpretation, what Kant is arguing in the transcendental deduction is that one of the necessary conditions of thinking about a given object is thinking that this object instantiates the categories. If we are thinking about a given object, we must be thinking that this object is a unity, or that it is a plurality, or that it is a totality, and so on for each of the other categories.Footnote ³ In this article I am going to focus on the second version of the non-conceptualist interpretation. However, as I shall explain in the fourth section of the article, the same basic concern that motivates the fitness-for-purpose objection can also be raised against the first version of the non-conceptualist interpretation. The proposal that I am going to make can therefore be understood as a general defence of the non-conceptualist interpretation of the transcendental deduction.

The clearest statement of the non-conceptualist interpretation in the recent secondary literature is due to Lucy Allais. She takes Kant to think that ‘intuitions do not depend on concepts to play their role of presenting us with particulars’ (2015: 148). According to her interpretation, the transcendental deduction is supposed to establish that ‘the application of the categories is necessary for us to have ‘relation to an object’ (A109) or for anything to ‘become an object for me’ (B138)’ (150). Allais insists that ‘we should be looking for an account of what Kant means by ‘relation to an object’ according to which it is something different from what he means by being given an object’ (268). Giving us objects is ‘the role of intuition and is something concepts can never do’ (268). For us to have ‘relation to an object’ is for us to have what Allais calls a ‘referential thought’ about an object that we have been given in intuition (150). It is for us to ‘apply concepts to this object’ (150). Allais concludes on this basis that the purpose of the transcendental deduction is to establish that the application of the categories to an object is a necessary condition of having a ‘referential thought’ about an object of empirical intuition. As she puts it, Kant ‘begins by assuming that we do have successful empirical concept application … [and] if he can show that applying a priori concepts to the objects presented in empirical intuition is a condition of applying empirical concepts to them, he will have shown that a priori concepts have relation to objects’ (270). Again, she writes that ‘[the] objective validity of the categories consists in their being used to think about objects that are given to us, and this objective validity is demonstrated by showing that using the categories is a condition of having any successful referential thought about such objects’ (271). If Allais is correct, what Kant is arguing in the transcendental deduction is that one of the necessary conditions of thinking about an object of empirical intuition is thinking that this object instantiates the categories.

Defenders of the non-conceptualist interpretation like to point to passages of the Critique that appear to suggest that what Kant is arguing in the transcendental deduction is precisely that the relation of a given object to the categories is a necessary condition of thinking about this object. Here are just a few examples:

There is also a well-known passage in which Kant appears to claim that the relation of a given object to the categories is not a necessary condition of having an empirical intuition of this object:

The standard response to this passage by defenders of the conceptualist interpretation is to suggest that Kant is simply pointing out that it is consistent with everything that he has argued so far in the Critique that the relation of a given object to the categories is not a necessary condition of having an empirical intuition of this object. According to this way of reading the passage, Kant is not suggesting that it is possible for us to have an empirical intuition of an object even if it is not related to the categories. He is simply alerting us to the fact that this possibility is one that he hasn’t yet ruled out. Defenders of the conceptualist interpretation think that ruling out this possibility is Kant’s first order of business in the rest of the transcendental deduction.Footnote ⁵

2

It is not my aim in this article to decide whether these passages provide a strong argument for the non-conceptualist interpretation. My focus is instead on a particular objection to this interpretation: the fitness-for-purpose objection. The objection has two premises. The first premise is that part of the purpose of the transcendental deduction is to refute a position that we might call, following Gomes (Reference Gomes2014: 11), Humean scepticism, without thereby meaning to ascribe this position to Hume himself. We can think of Humean scepticism as coming in two forms, corresponding to the two forms of idealism that Kant distinguishes at the beginning of the Refutation of Idealism (B274). According to extreme Humean scepticism, there are no objects of empirical intuition that instantiate the categories. None of the objects of empirical intuition is a unity, none of them is a plurality and none of them is a totality. Nor is any of these objects a reality, a negation or a limitation, and so on for the rest of the categories. According to the more moderate form of Humean scepticism, we simply do not know that there are objects of empirical intuition that instantiate the categories. We do not know that any of the objects of empirical intuition is a unity, or that any of these objects is a plurality, or a totality. Nor do we know that any of the objects of empirical intuition is a reality, a negation or a limitation, and so on for the rest of the categories.

It is important to emphasize that this is a weak assumption about the aim of the transcendental deduction. It is perfectly compatible with the idea that Kant intends to refute forms of scepticism other than Humean scepticism, or that he intends to establish important claims about the necessary conditions of thought and empirical intuition. The assumption upon which the fitness-for-purpose objection rests is just that if the transcendental deduction fails to refute Humean scepticism then it has not done one of the things that it is supposed to do.Footnote ⁶

The second premise of the objection is a simple logical point. Suppose that the non-conceptualist interpretation is correct. Suppose that what Kant is arguing in the transcendental deduction is that one of the necessary conditions of thinking about a given object is thinking that it instantiates the categories. Suppose we also take it for granted that there are objects of empirical intuition that we think about. Then it certainly follows from what Kant is arguing in the transcendental deduction that there are objects of empirical intuition that we think instantiate the categories. But it does not follow from what Kant is arguing that there are objects of empirical intuition that instantiate the categories. And since it does not follow from what Kant is arguing that there are objects of empirical intuition that instantiate the categories, it does not follow from what Kant is arguing that we know that there are objects of empirical intuition that instantiate the categories. If the non-conceptualist interpretation is correct, what Kant is arguing in the transcendental deduction is perfectly consistent with both the extreme and more moderate forms of Humean scepticism. It could be the case that there are no objects of empirical intuition that instantiate the categories, in spite of the fact that one of the necessary conditions of thinking about a given object is thinking that this object instantiates the categories. Quoting Gomes:

The conclusion of the objection is that if the non-conceptualist interpretation is correct, the transcendental deduction does not do what it is supposed to do. Part of the purpose of the transcendental deduction is to refute Humean scepticism. But if the non-conceptualist interpretation is correct, what Kant is arguing in the transcendental deduction is perfectly consistent with Humean scepticism. So if the non-conceptualist interpretation is correct, the transcendental deduction is not fit for purpose.Footnote ⁷

It may be asked why I have presented the fitness-for-purpose objection as an objection to the non-conceptualist interpretation in particular. Notice that we can distinguish two aspects of this interpretation. First, there is the fact that it takes Kant to be specifying one of the necessary conditions of thinking about a given object, as opposed to specifying one of the necessary conditions of having an empirical intuition of a given object. Second, there is the fact that the interpretation takes the necessary condition in question to be thinking that the object instantiates the categories. Now according to the way the debate is usually set up, it is the first aspect of this interpretation that makes it non-conceptualist. Yet it is the second aspect of the interpretation that is the focus of the fitness-for-purpose objection. It is perfectly possible to come up with a conceptualist interpretation of the transcendental deduction that is also open to the fitness-for-purpose objection, namely, one that takes Kant to be arguing that one of the necessary conditions of having an empirical intuition of a given object is thinking that this object instantiates the categories. This sort of conceptualist interpretation shares the second aspect of the non-conceptualist interpretation, but not the first. It is worth noting that it is precisely because James Van Cleve accepts this sort of conceptualist interpretation that he thinks that the transcendental deduction is not fit for purpose (Van Cleve Reference Van Cleve1999: 79–89).

The reason that I have presented the fitness-for-purpose objection as an objection to the non-conceptualist interpretation is that I do not think that the sort of conceptualist interpretation that I just mentioned is a serious candidate in debates about the transcendental deduction.Footnote ⁸ It is extremely difficult to square with Kant’s claims in the introduction to the Transcendental Logic about the importance of not confusing the different roles of sensibility and the understanding (A51–2/B76). Here Kant writes that ‘[w]ithout sensibility no object would be given to us, and without understanding none would be thought’ (A51/B75). Yet if the sort of conceptualist interpretation that I just mentioned is correct, Kant should have said that without understanding no object would be given to us. If the conceptualist interpretation is correct, and Kant thinks that the application of the categories to a given object is a necessary condition of having an empirical intuition of this object, what he means by the application of the categories to this object must be something other than thinking that this object instantiates the categories.Footnote ⁹

3

Before I consider some possible responses to the fitness-for-purpose objection, I want to distinguish it from two other objections with which it might easily be confused. The first of these is what I am going to call the triviality objection. According to the fitness-for-purpose objection, the problem with the non-conceptualist interpretation is that, if it is correct, the transcendental deduction does not do what it is supposed to do. It does not refute Humean scepticism. According to the triviality objection, the problem with the non-conceptualist interpretation is that, if it is correct, the transcendental deduction does not establish anything of philosophical interest or importance and so does not do anything worth doing.

One philosopher who has raised the triviality objection against the non-conceptualist interpretation is Hannah Ginsborg. She suggests that this interpretation ‘threatens to trivialize Kant’s central project in the Critique, or at least to diminish its interest and importance’ (Reference Ginsborg2006: 62). She also writes that

The trouble with the triviality objection is that it is based on a false premise. If the transcendental deduction establishes that one of the necessary conditions of thinking about a given object is thinking that this object instantiates the categories, it does establish something of philosophical interest and importance.Footnote ¹⁰ It is not trivial that there is a special group of concepts such that one of the necessary conditions of thinking about a given object is thinking that it instantiates these concepts. Nor is it trivial that if there is a special group of concepts that satisfies this description, then it is the one presented in Kant’s table of the categories.Footnote ¹¹

Ginsborg mentions in a footnote that this response has been put to her by Eric Watkins.Footnote ¹² Her counter-response is that, while ‘perhaps “triviality” is too strong a term’, ‘it has to be conceded that Kant’s conclusion, understood in the light of this approach [i.e. the non-conceptualist interpretation], is considerably less significant than might be hoped’ (2006: 99). Since it is not clear on what grounds this might be hoped other than the evidence that establishing a more significant conclusion is part of the purpose of the transcendental deduction, Ginsborg appears here to be leaning away from the triviality objection and towards the fitness-for-purpose objection. She puts her support squarely behind the fitness-for-purpose objection in a more recent article. Here she describes the idea that the categories have ‘a role to play, not just in explicit judgment, but also in our perceptual apprehension of the objects about which we judge’ as ‘essential to the anti-Humean aspect of Kant’s view in the Critique’ (2008: 70). She adds that the non-conceptualist interpretation ‘seems to leave Kant without a response to the Humean worry … that because objects of sensible intuition might not conform to the conditions of the synthetic unity of thought, the concept of cause might be “empty, null and meaningless” (A90/B122)’ (2008: 70).

The second objection from which I want to distinguish the fitness-for-purpose objection is what I am going to call the rogue objects problem. This is the subject of a recent exchange between Robert Hanna (Reference Hanna2011) and Stefanie Grüne (Reference Grüne2011). It has the same structure and conclusion as the fitness-for-purpose objection. However, its specific target is the first version of non-conceptualist interpretation mentioned in section 1. According to this interpretation, what Kant is arguing in the transcendental deduction is that a necessary condition of thinking about a given object is the instantiation by this object of the categories. The first premise of the rogue objects problem is that part of the purpose of the transcendental deduction is to rule out the possibility of ‘rogue objects’, where this means objects of empirical intuition that do not instantiate the categories. The second premise of the rogue objects problem is that if the first version of the non-conceptualist interpretation is correct, the transcendental deduction fails to rule out the possibility of rogue objects. Suppose that what Kant is arguing in the transcendental deduction is that one of the necessary conditions of thinking about a given object is the instantiation by this object of the categories. Then it certainly follows from what Kant is arguing in the transcendental deduction that there are no objects of empirical intuition that we think about and that do not instantiate the categories. But it does not follow from what Kant is arguing that there are no objects of empirical intuition that do not instantiate the categories, and so it does not follow that there are no rogue objects. It only follows that there are no rogue objects if we take it for granted that there are no objects of empirical intuition that we do not think about, and this is not something that we can plausibly take for granted. The conclusion of the rogue objects problem is that if the first version of the non-conceptualist interpretation is correct, the transcendental deduction is not fit for purpose.

The crucial point to bear in mind when comparing the fitness-for-purpose objection to the rogue objects problem is that while both of these objections suggest that, according to the non-conceptualist interpretation, there is a logical gap between what the transcendental deduction establishes and what it is supposed to establish, the logical gap is different in each case. According to the fitness-for-purpose objection, the logical gap is between the claim that there are objects of empirical intuition that we think instantiate the categories and the claim that there are objects of empirical intuition that instantiate the categories. According to the rogue objects problem, the logical gap is between the claim that there are no objects of empirical intuition that we think about and that do not instantiate the categories, and the claim that there are no objects of empirical intuition that do not instantiate the categories.

4

One possible response to the fitness-for-purpose objection is to bite the bullet. We might accept that the fitness-for-purpose objection gives us a reason to reject the non-conceptualist interpretation. But we might also insist that this is outweighed by the numerous reasons to accept the non-conceptualist interpretation. We would then be forced to accept that the transcendental deduction is not fit for purpose.

But there is a serious problem with this response. If the transcendental deduction is not fit for purpose in the way that I explained in the second section of this article, there is a huge logical gap at the heart of the Critique. Kant has concluded that there are objects of empirical intuition that instantiate the categories on the grounds that, first, there are objects of empirical intuition that we think about, and second, a necessary condition of thinking about a given object is thinking that it instantiates the categories. If anything counts as a ‘non sequitur of numbing grossness’ (Strawson Reference Strawson1966: 137), this does. As we might also put it, paraphrasing Woozley (Reference Woozley1964: 26–7), it would be hard to understand why anybody should want to rate Kant an important philosopher if his whole theory rested on errors so elementary that a first-year student would have no difficulty in spotting them. That Kant would have failed to spot this logical gap is just implausible.

Another possible response to the fitness-for-purpose objection is to reject the first premise of the objection, which is that part of the purpose of the transcendental deduction is to refute Humean scepticism. It may be suggested that the transcendental deduction has a more modest goal, which it can accomplish even if the non-conceptualist interpretation is correct. Gomes considers a version of this response according to which it is no part of the purpose of the transcendental deduction to refute Humean scepticism, but it is part of the purpose of the transcendental deduction to refute a different sort of scepticism, which ‘[raises] the question of how it is that a priori concepts can be applied to experience, not whether they can ever be accurately applied so’ (2014: 14). Whereas a Humean sceptic thinks that that there are no objects of empirical intuition that instantiate the categories (or that we do not know that there are objects of empirical intuition that instantiate the categories), a sceptic of this other sort thinks that there are no objects of empirical intuition that we think instantiate the categories (or that we do not know that there are objects of empirical intuition that we think instantiate the categories). Now it is easy to see how the transcendental deduction refutes this sort of scepticism if the non-conceptualist interpretation is correct. Suppose that what Kant is arguing in the transcendental deduction is that a necessary condition of thinking about a given object is thinking that it instantiates the categories. And suppose we take it for granted that there are objects of empirical intuition that we think about. Then it follows from what Kant is arguing in the transcendental deduction that there are objects of empirical intuition that we think instantiate the categories.

There is the occasional passage in Allais’ discussion that suggests this sort of response to the fitness-for-purpose objection. I have already quoted her remark that ‘[the] objective validity of the categories consists in their being used to think about objects that are given to us’ (2015: 271). What this suggests is that the categories are objectively valid if and only if there are objects of empirical intuition that we think instantiate the categories. So to the extent that the purpose of the transcendental deduction is to establish that the categories are objectively valid, the purpose of the transcendental deduction is to establish that there are objects of empirical intuition that we think instantiate the categories. But there are also some problems with this response to the fitness-for-purpose objection. One is that it deals with the objection only at the cost of diminishing the philosophical interest and importance of the transcendental deduction. Take the cup of tea on the desk in front of me. It is an object of empirical intuition, and it is also an object that I think instantiates the categories. It is an object that I think actually exists. And it is an object that I think stands in relations of causality. So there is an object of empirical intuition that I think instantiates the categories. There is no need for a lengthy, complicated refutation of the sort of scepticism according to which there are no objects of empirical intuition that we think instantiate the categories. This sort of scepticism is trivially false. To the extent that the purpose of the transcendental deduction is to refute this sort of scepticism, it does not do anything worth doing.

Another problem with the present response is that it is hard to square with the text of the Critique. First, there are passages that strongly suggest that the purpose of the transcendental deduction is to justify our application of the categories to objects of empirical intuition. Kant opens the transcendental deduction by introducing the legal concept of a deduction as a proof the point of which is ‘to demonstrate [an] entitlement or legal claim’ (A84/B116, my emphasis). He then writes that while ‘we help ourselves to a multitude of empirical concepts without objection from anyone and take ourselves to be justified … in granting them a sense and imagined meaning’ (A84/B116, my emphasis), there are ‘among the many concepts that constitute the mixed fabric of human cognition … some also that are intended for pure use a priori … and the entitlement of these always requires a deduction’ (A85/B117, my emphasis). Now to justify our application of the categories to objects of empirical intuition, it is not sufficient to establish that there are objects of empirical intuition that we think instantiate the categories. The question whether or not our application of the categories to objects of empirical intuition is justified only arises on the assumption that there are objects of empirical intuition that we think instantiate the categories. Nor is it sufficient to establish that our application of the categories to objects of empirical intuition is necessary. This might show that we were not to blame for our application of the categories to objects of empirical intuition, but it would not justify it. For this, it is necessary to establish that, when we apply the categories to objects of empirical intuition, we are, at least sometimes, getting it right. Thus the transcendental deduction justifies our application of the categories to objects of empirical intuition only if it refutes Humean scepticism.

There are other passages of the Critique that suggest in a more direct way that it is part of the purpose of the transcendental deduction to refute Humean scepticism. Consider, for example, the famous passage where Kant uses the example of the concept of cause and effect to illustrate the problem that he wants to solve:

Here Kant indicates that the purpose of the transcendental deduction is to establish that the categories are not empty and that they find objects among the appearances. This is naturally interpreted as the claim that the purpose of the transcendental deduction is to establish that there are appearances – objects of empirical intuition – that instantiate the categories.

A third possible response to the fitness-for-purpose objection is to abandon the second version of the non-conceptualist interpretation that I introduced in section 1 and to plump instead for the first version. According to this version of the non-conceptualist interpretation, remember, what Kant is arguing in the transcendental deduction is not that one of the necessary conditions of thinking about a given object is thinking that this object instantiates the categories, but rather that one of the necessary conditions of thinking about a given object is the instantiation of the categories by this object. If this interpretation is correct, it certainly follows from what Kant is arguing in the transcendental deduction that there are objects of empirical intuition that instantiate the categories, as long as we take it for granted that there are objects of empirical intuition that we think about.

One problem with this response is that the alternative non-conceptualist interpretation is open to the rogue objects problem that I mentioned in section 3. Another problem is that the logical gap to which the fitness-for-purpose objection draws attention now simply threatens to crop up in the middle of the argument of the transcendental deduction.Footnote ¹³ On the first version of the non-conceptualist interpretation, there must still come a point where Kant has to move from a claim about the necessary conditions of thinking about a given object to the claim that this object instantiates the categories. It is not clear how Kant can bridge the logical gap between these claims, and if he fails to bridge this gap then the transcendental deduction is still unfit for purpose.Footnote ¹⁴ In this way, the same basic concern that motivates the fitness-for-purpose objection can also be raised in relation to the first version of the non-conceptualist interpretation. To be clear, my point is not that there is no way for a defender of the first version of the non-conceptualist interpretation to respond to this concern. My point is that this concern is not addressed by moving from the second version of the non-conceptualist interpretation to the first.

5

There is a better response to the fitness-for-purpose objection available to defenders of the non-conceptualist interpretation. The first stage in this response is to make the following point. If the non-conceptualist interpretation is correct, what Kant is doing in the transcendental deduction is defending a claim about the necessary conditions of thinking about a given object. Now we might reasonably assume that one of the things that counts as thinking about a given object, say, the cup of tea on the desk in front of me, is thinking that this object does not instantiate the categories – thinking that this cup of tea is not a unity, a plurality or a totality, a reality, a negation or a limitation, and so on for the other categories. We might also assume that one of the things that counts as thinking about a given object such as this cup of tea is thinking that it is something that we do not know instantiates the categories. So if the non-conceptualist interpretation is correct, it follows from what Kant is arguing in the transcendental deduction that a necessary condition of thinking that a given object does not instantiate the categories, and of thinking that we do not know that a given object instantiates the categories, is thinking that this object does instantiate the categories.

Now if I think both that a given object instantiates the categories and that this object does not instantiate the categories, I am thinking straightforwardly contradictory things. If I think both that a given object instantiates the categories and that we don’t know that this object instantiates the categories, what I am thinking is, if not a straightforward contradiction, at least Moore-paradoxical.Footnote ¹⁵ Generally, if the non-conceptualist interpretation is correct, it follows from what Kant is arguing in the transcendental deduction that a necessary condition of thinking that a given object does not instantiate the categories, or that we do not know that a given object instantiates the categories, is having thoughts that are in some way or another incoherent. In that sense, it follows from what Kant is arguing in the transcendental deduction that I cannot coherently think that a given object does not instantiate the categories, and I cannot coherently think that we do not know that this object instantiates the categories. For I cannot think these things without also thinking that the object does instantiate the categories.

The second stage in the present response is to link this point to Humean scepticism. One example of thinking that a given object does not instantiate the categories is thinking that none of the objects of empirical intuition instantiates the categories, and one example of thinking that we do not know that a given object instantiates the categories is thinking that we do not know that any of the objects of empirical intuition instantiates the categories. The given objects in these examples are the objects of empirical intuition in general. Putting this suggestion together with the first stage of the response, we can draw the following conclusion. If the non-conceptualist interpretation is correct, it follows from what Kant is arguing in the transcendental deduction that Humean scepticism is not something that we can coherently think is true. We cannot coherently think that none of the objects of empirical intuition instantiates the categories. Nor can we coherently think that we do not know that any of the objects of empirical intuition instantiates the categories. We cannot think either of these things without also thinking that the objects of empirical intuition in general instantiate the categories.

The final step in the present response is to suggest that this amounts to a refutation of Humean scepticism. The fitness-for-purpose objection turns on the assumption that the transcendental deduction refutes Humean scepticism only if it follows from what Kant is arguing that Humean scepticism is false, which is to say, only if it follows that there are objects of empirical intuition that instantiate the categories, and, moreover, that we know that there are objects of empirical intuition that instantiate the categories. But that isn’t obviously true. Kant may well think that the transcendental deduction refutes Humean scepticism if it follows from what he is arguing that Humean scepticism is not something that we can coherently think is true.Footnote ¹⁶

If this response to the fitness-for-purpose objection is correct, the transcendental deduction adopts a version of what Robert Stern has called the ‘modest transcendental argument strategy’ (Reference Stern2008: 268).Footnote ¹⁷ But it is important to distinguish different senses in which a transcendental argument might be described as ‘modest’. According to the present response, the transcendental deduction is not modest in the sense that it gives up on the goal of justifying our belief that there are objects of empirical intuition that instantiate the categories. It is not modest in the sense that it adopts the more limited goal of showing that this belief is something we cannot do without. Rather, it is modest in the sense that it gives up on the goal of showing it to be true that there are objects of empirical intuition that instantiate the categories and adopts instead the more limited goal of showing this to be something that we cannot coherently doubt or deny.Footnote ¹⁸

This interpretation helps us to make sense of important passages of the first Critique. For example, in the Preface to the first edition, regarding the part of the transcendental deduction that is supposed to ‘demonstrate and make comprehensible the objective validity of [the understanding’s] concepts a priori’ (Axvi), Kant writes that ‘what is said at pages 92 to 93 can be sufficient by itself’ (Axvii). Now what Kant actually does at A92 and 93 is to distinguish the conditions of the possibility of intuition from the conditions of the possibility of thinking about objects, and to raise the possibility that the pure concepts of the understanding are conditions of the possibility of thinking about objects. He writes that ‘the objective validity of the categories … rests on the fact that through them alone is experience possible (according to the form of thinking)’ (my emphasis), and the categories ‘are thereupon related necessarily and a priori to objects of experience, because only by means of them can any object of experience at all be thought’ (A93/B126, my emphasis). This passage strongly suggests that, according to Kant, it is sufficient to refute Humean scepticism to prove that the application of the categories is a necessary condition of thinking about an object. Even defenders of the conceptualist interpretation need to explain why Kant thinks that this is true. My response to the fitness-for-purpose objection provides a plausible answer. According to Kant, if the application of the categories is a necessary condition of thinking about an object, then even raising the question of the relation of the categories to objects involves thinking of these objects as instantiating the categories.

There are some possible objections to the present response. First, it may be suggested that there is another version of the fitness-for-purpose objection to which the non-conceptualist interpretation is still open. Suppose that the non-conceptualist interpretation is correct. Then it follows from what Kant is arguing in the transcendental deduction that a necessary condition of thinking that the objects of empirical intuition do not instantiate the categories is thinking that the objects of empirical intuition instantiate the categories. Suppose it also follows from this that we cannot coherently think that the objects of empirical intuition do not instantiate the categories. From this it certainly follows that there are objects of empirical intuition that we cannot coherently think do not instantiate the categories. But it does not follow that there are objects of empirical intuition that instantiate the categories. So the transcendental deduction is still not fit for purpose. It fails to rule out the possibility that we are subject to an illusion whereby we cannot think about the objects of empirical intuition without thinking that they instantiate the categories, and whereby we cannot even coherently think that the objects of empirical intuition do not instantiate the categories, in spite of the fact that the objects of empirical intuition do not instantiate the categories. This illusion is even more pernicious than the one described by Gomes (Reference Gomes2014: 11), since it involves the inability coherently to think that it is an illusion.Footnote ¹⁹

But consider what exactly is being proposed: that it might be true both that we cannot coherently think that the objects of empirical intuition do not instantiate the categories and that the objects of empirical intuition do not instantiate the categories. This proposal is something that we can coherently think is true only if each of its conjuncts is something that we can coherently think is true. But if the first conjunct of this proposal is true, the second conjunct is not something that we can coherently think is true. So this proposal is something that we can coherently think is true only if its first conjunct is not true. And if its first conjunct is not true, then neither is the conjunction. So if the proposal is something that we can coherently think is true, it is not true. Conversely, if it is true, it is not something that we can coherently think is true.

I suggest that the temptation to object to the present response to the fitness-for-purpose objection in this way is based on a failure to take on board the point that Kant is making in the transcendental deduction according to the non-conceptualist interpretation. In raising this objection, we take ourselves to be able coherently to entertain the possibility that everything Kant is arguing in the transcendental deduction is true, in spite of the fact that the objects of empirical intuition do not instantiate the categories. But if everything Kant is arguing in the transcendental deduction is true, this is not a possibility that we can coherently entertain. So in raising this objection, we are either failing to realize that the possibility we are entertaining is one in which this possibility is one that we cannot coherently entertain, or we are simply insisting, contrary to what Kant is arguing in the transcendental deduction, that this is a possibility that we can coherently entertain.

The second way of objecting to the present response that I want to mention in this article involves rejecting the assumption that thinking that Humean scepticism is true is an example of thinking about a given object – having what Allais calls a ‘referential thought’ – in the sense at issue in the transcendental deduction. The point is strongest in the case of the more moderate form of Humean scepticism. It may be suggested that while Kant certainly thinks that a necessary condition of thinking that the cup of tea I can see on the table in front of me does not instantiate the categories is thinking that it does instantiate the categories – since this thought is clearly a thought about a given object – Kant does not think that a necessary condition of thinking that we do not know that there are objects of empirical intuition that instantiate the categories is thinking that objects of empirical intuition instantiate the categories. This is not a thought about objects of empirical intuition. On the contrary, it is a thought about what we do and do not know about objects, and it does not follow from what Kant is arguing in the transcendental deduction according to the non-conceptualist interpretation that the application of the categories is a necessary condition of thinking about what we do and do not know about objects.

The key to this objection is the assumption that, as far as Kant is concerned, the thought that we do not know that there are objects of empirical intuition that instantiate the categories is not a thought about the objects of empirical intuition in general, but rather a thought about something else entirely, namely, ourselves and our mental states. But there is no need for Kant to deny that the thought that we do not know that there are objects of empirical intuition that instantiate the categories is a thought about the objects of empirical intuition in general. He could simply claim that this thought is indirectly about the objects of empirical intuition in general, whereas the thought that there are no objects of empirical intuition that instantiate the categories is directly about the objects of empirical intuition in general. There is a clear sense in which the first thought is about the objects of empirical intuition in general. In having this thought, we are thinking about the objects of empirical intuition in general, as well as about ourselves and our mental states. So it plausibly follows from the conclusion of the transcendental deduction that a necessary condition of having this thought is thinking that the objects of empirical intuition in general instantiate the categories.

A third way of objecting to my proposal is to suggest that there is no point defending the non-conceptualist interpretation against the fitness-for-purpose objection in light of the textual evidence against this interpretation, and, in particular, the evidence that the transcendental deduction turns on what I am going to call the synthesis argument. The first premise of this argument is that the application of the categories is a necessary condition of synthesis, and the second premise is that synthesis is a necessary condition of the empirical intuition of an object. The conclusion is that the application of the categories is a necessary condition of the empirical intuition of an object. One way for a non-conceptualist to respond to this objection is to question the evidence that Kant accepts the first premise of the synthesis argument. Another way is to point out that Kant distinguishes different types of synthesis (e.g. B151–2), and to suggest that the type of synthesis for which the first premise is true is different to the type of synthesis for which the second premise true. I follow Lucy Allais (Reference Allais2017) and Colin McLear (Reference McLear2015) in questioning the evidence that Kant accepts the second premise of the argument. In the first edition transcendental deduction, Kant writes that ‘for unity of intuition to come from this manifold … the running-through and then taking-together of this manifoldness is necessary, which action I call the synthesis of apprehension’ (A99). He adds that this synthesis is ‘directed straight at the intuition’ (A99). This may be taken to mean that the synthesis of apprehension is the act of unifying various representations so as to produce an intuition. However, it is also plausibly interpreted as the claim that the synthesis of apprehension is the act of unifying various intuitions so as to produce a higher-order representation. This second interpretation is supported by Kant’s claim in the second edition transcendental deduction that the synthesis of apprehension is ‘the composition of the manifold in an empirical intuition, through which perception … becomes possible’ (B160), which suggests that the synthesis of apprehension presupposes empirical intuition. Now in the same passage Kant does say that ‘all synthesis, through which even perception becomes possible, stands under the categories’ (B161), which implies that synthesis is a necessary condition of perception. However, as McLear has pointed out (2015: 101), it is clear that the term ‘perception’ is not being used at this point as a synonym for ‘empirical intuition’. Again, the famous footnote to this passage is taken by defenders of the conceptualist interpretation to express the claim that synthesis is a necessary condition of the intuitions of space and time (e.g. Longuenesse Reference Longuenesse1998: 213, 225). But what Kant writes in this footnote is that a particular unity of representation ‘presupposes a synthesis, which does not belong to the senses, through which all concepts of space and time first become possible’ (B160–1n., my emphasis).Footnote ²⁰ While these are controversial passages, it is perfectly reasonable to interpret them as expressing something other than the second premise of the synthesis argument. They do not provide conclusive evidence for the conceptualist interpretation of the transcendental deduction.

Finally, it may be objected that the position that I have defended in this article fails to address other problems typically associated with the non-conceptualist interpretation of the transcendental deduction. One example is the redundancy objection. According to the fitness-for-purpose objection, the trouble with the non-conceptualist interpretation is that, if it is correct, then the transcendental deduction does not do what it is supposed to do. According to the redundancy objection, the trouble with the non-conceptualist interpretation is that, if it is correct, then the transcendental deduction does not do anything that has not already been done in the metaphysical deduction, where Kant argues that one of the necessary conditions of thinking about a given object is thinking that it instantiates the categories. So if this is also what Kant is arguing in the transcendental deduction, it is redundant.

Another important objection to the non-conceptualist interpretation is the two-steps problem. According to this objection, the trouble with the non-conceptualist interpretation is that, if it is correct, then it is difficult to explain the two-step structure of the transcendental deduction in the second edition of the Critique – the division between §§15–20, where ‘the beginning of a deduction of the pure concepts of the understanding is made’, but where it is necessary to abstract ‘from the way in which the manifold is given for an empirical intuition’ (B144), and §§22–6, where ‘from the way in which the empirical intuition is given in sensibility, it [is] shown that the unity [of this intuition] is none other than that which the category prescribes to the manifold of a given intuition in general’ (B145).Footnote ²¹

A comprehensive response to either of these objections on behalf of the non-conceptualist interpretation lies beyond the scope of this article. However, I can indicate plausible lines of response. Regarding the two-steps problem, I suggest that while Kant’s argument up to §20 is supposed to establish that the application of the categories is a necessary condition of thinking about an object, his argument in §26 is supposed to reach the further conclusion that the application of the categories is a necessary condition of unifying intuitions. It does this by establishing that thinking is a necessary condition of unifying intuitions – and, in particular, by ruling out the possibility of an alternative way of unifying intuitions that exploits the unity of space and time. An interpretation along these lines makes it possible to understand the second step of the second edition transcendental deduction as establishing something more than the claim that the application of the categories is a necessary condition of thinking about an object, without seeing it as establishing something incompatible with non-conceptualism.

Regarding the redundancy objection, Kant makes it clear in the Preface to the first edition of the Critique that his aim in the transcendental deduction is not only to prove that the application of the categories is a necessary condition of thinking about an object, but also to explain why the application of the categories is a necessary condition of thinking about an object, by addressing the question ‘how is the faculty of thinking itself possible?’ (Axvii). So even if the non-conceptualist interpretation is correct, it does not follow that the transcendental deduction is redundant in light of what Kant has already established in the metaphysical deduction. He may have come close to achieving the first aim of the transcendental deduction by the end of the metaphysical deduction, but he hasn’t even started to address the second.

6

In this article I have examined the fitness-for-purpose objection against the non-conceptualist interpretation of Kant’s transcendental deduction. The objection turns on the assumption that if the non-conceptualist interpretation is correct, then the transcendental deduction fails to refute Humean scepticism. I have argued that this assumption is false. If the non-conceptualist interpretation is correct, it does not follow from what Kant is arguing in the transcendental deduction that Humean scepticism is false. However, it does plausibly follow from what Kant is arguing that Humean scepticism is not something that we can coherently think is true. Kant may well think that this is sufficient to refute Humean scepticism.Footnote ²¹