2. Generation (of the) Gap

In my (2010), I set out the potential problem for interpretations of the transcendental deduction using Van Cleve’s distinction between the claim that we must apply the categories to the objects of experience and the claim that the categories apply to the objects of experience. In a recent discussion of Lucy Allais’s Manifest Reality (Allais Reference Allais2015) I phrased this as a distinction between a claim about the application of the categories and claim about their instantiation (Gomes Reference Gomes2017b).

Schulting is somewhat wary of these formulations, and instead settles on a gap between:

and

Part of this is likely terminological: in my idiolect ‘instantiation’ is a relation between a concept and its instances, such as to talk about the categories being instantiated in the perception of objects would be to claim that the objects of perception were themselves categorial. Schulting, I take it, uses ‘exemplification’ for this relation. The terminology does not matter. What is important is that there is a potential gap between a claim about the subject – that she judges, experiences, perceives or intuits objects as subject to the categories – and a claim about the object – that the objects of judgements, experience, perception or intuition sometimes exemplify the categories. This is the gap that one might think needs to be bridged.

One point is more substantial: Schulting takes the destination claim to be one about the objects of experience necessarily exemplifying the categories. This seems too strong since, on the face of it, it looks like there can be categorial illusions: cases where the objects of judgement, experience or perception seem to exemplify some category or other but actually fail to do so. It is not clear to me whether Schulting thinks this is possible: in a very interesting discussion of the nature of objective validity, he claims that ‘objective validity is not a merely logical condition in the sense of a judgement’s being either true or false’, but rather concerns ‘what Kant calls transcendental truth (B185/A146)’ (p. 154), and that the ‘categories are not only necessary for objective experience, but also, formally at least, sufficient for it’ (p. 146). If this is a point about the subject side of the gap – witness the claim about objective experience in the previous quote – then I am sympathetic to Schulting’s point. But if this is meant to be a claim about the object side – that necessarily the categories are exemplified by the objects of our judgement, experience or perception – then it would be interesting to hear whether and how this is supposed to be compatible with the possibility of categorial illusions.

I will ignore this complication and use Schulting’s modally stronger formulation in what follows though, for my part, I would prefer to weaken the destination claim to one about possibility. Still with this in mind we can phrase the issue generically as a gap between:

and

The variable ‘ϕ’ can be replaced by some activity on the part of the subject, be it judging, experiencing, perceiving or intuiting. And the gap is to be understood as the distinction between (S) and (O).

Noting that ‘ϕ’ is a variable which can be filled in in different ways is important, since it makes explicit that differing interpretations can disagree as to the activity in question. This matters because the choice of activity may make a difference to whether it is plausible to think there is a gap here to be filled. One example, as an aside, to make this point: Lucy Allais’s interpretation of the transcendental deduction takes the activity in question to be our thinking of objects in some way. Schulting claims against Allais that this interpretation makes Kant’s argument ‘easily vulnerable to the Gap’ (p. 184, n. 5). I agree (Gomes Reference Gomes2017b). So the question of whether there is a gap between (S) and (O) cannot be separated from questions about the epistemic capacities in question.

Structuring this debate as concerning Kant’s grounds for moving from (S) to (O) is helpful because it makes clear that this is effectively an iteration or an instance of the debate between Stroud (Reference Stroud1968) and Strawson (Reference Strawson1966) concerning the structure and viability of transcendental arguments. The issue there, as here, is one about what grounds there are for moving from claims about the capacities of subjects to claims about the nature of objects. And of course this issue traces back to the very first engagements with the first Critique. So the issues here run deep.

When structured as above, we can see that there are four options for responding to any alleged gap:

1. (1) There is no gap to be bridged because the transcendental deduction only aims at establishing (S).

2. (2) There is no gap to be bridged because the transcendental deduction establishes (O) without going via (S).

3. (3) There is no gap to be bridged because in establishing (S) Kant ipso facto establishes (O).

4. (4) There is a gap to be bridged between (S) and (O).

Schulting, as I understand him, holds (3). The fourth option comes in two flavours. Van Cleve seems to hold that there is a gap to be bridged but that Kant fails to bridge it; whereas in my (2010) I argued that there is a gap to be bridged but that Kant attempted to bridge it in the second part of the B-Deduction. I turn now to Schulting’s criticisms of my attempt on Kant’s part to bridge that gap.

3. A Credibility Gap?

In my (2010) I argued that the conclusions of the two parts of the B-Deduction correspond to some version of (S) and (O) above. That is, the first part of the B-Deduction (§§15–19) establishes a claim about the epistemic capacities of rational subjects: that we must experience or perceive objects as falling under the categories. And the second part of the B-Deduction (§§21–6) completes the argument by showing not only that we must experience or perceive objects as falling under the categories, but also that the categories must be capable of being exemplified by the objects of experience and perception.

Schulting raises three problems with this approach. They turn on the details of how I suggest Kant argues for the second step. My views have changed since the (2010) paper, so what follows is a sketch of what someone who liked that kind of view might say in response to Schulting’s criticisms. Here is how he presents the view of the (2010) paper:

And against this he raises three problems. First, that ‘there is an interpretative problem for Gomes’s proposal. On Kant’s view, objects do not necessitate the instantiation of the categories, as if they were somehow disposed to do so’. Second, that ‘if indeed the objects themselves necessitated their exemplification and it is not the subject that establishes the exemplification, how could I know that the categories are exemplified in an object?’. And third, that ‘there is a structural problem with Gomes’s proposal: … if there is a fundamental problem with the “first step”, namely … the gap between (N1*) and (N3), then the “second step” cannot solve it’ (from pp. 159–60).

The issues here are somewhat tangential to Schulting’s main line of argument in his chapter, so let me just sketch the kind of response that someone might give to these objections. Part of it involves getting clear on the proposed solution. As I understood it, this response to the gap turns on a particular understanding of what is going on in §§24 and 26 of the B-Deduction. The central claim is that the unity of space and time presuppose a synthesis in which ‘the understanding determines the sensibility’ (B161n.), and it is this transcendental synthesis which explains the fact that what is presented in space and time is such as to require synthesis in accordance with the categories. However, Kant also tells us that this synthesis is one which ‘precedes all concepts’ (B161n.), so we cannot think of the synthesis responsible for the unity of space and time as one which involves the categories: the synthesis in question is undertaken by the understanding, but without any use of concepts.

One might think that this suggestion – that there is a use for the understanding which does not involve synthesis in accordance with concepts – is obviously absurd, conflicting as it does with Kant’s claim that the understanding is the ‘faculty for judging’ (A69/B94). But a defender of this reading will hold that there is a more fundamental characterisation of the understanding as the capacity for apperception (as Kant suggests at B133–4n.), and that this allows the possibility of a non-conceptual use of the understanding. It is this aspect of the understanding – the understanding as the unity of apperception – which is responsible for the unity of space and time. To use more recent terminology (from McLear Reference McLear2015), this reading takes an intellectualist but non-conceptualist approach to the unity of space and time to be: the unity of space and time depends on the activity of the understanding, but not on any concept-governed synthesis (cf. Land Reference Land2006; Longuenesse Reference Longuenesse1998).

How does this reading of the second part of the B-Deduction help bridge the gap between (S) and (O)? The idea is that although the unity of space and time does not itself involve the categories, it has its origins in the same source as the categorial synthesis which takes place on the manifold of intuition given in space and time – namely, in the understanding understood as the synthetic unity of apperception. And it is in virtue of sharing this origin that what is given in space and time is such as to require synthesis in accordance with the rules by which the understanding imposes unity, which is to say, the categories.

This sketch needs filling out in a number of ways, but I think it would be misleading to say of this proposal that it takes the categories to be ‘the very rules by means of which objects are given in space’ (p. 158), for the categories do not enter the picture until the understanding operates discursively on that which is given to us in space and time. So it is not that ‘objects themselves are already given as categorially governed particulars in a spatial continuum and are as such apprehended by the experiencing and category-applying subject’ (p. 158), but that objects are given as falling under the unity of apperception in such a way that they are suitable for unification under the categories.

This may help undercut the first and second of Schulting’s worries. On the first worry, Schulting objects that ‘objects cannot necessitate the instantiation of the categories, for, as Kant says, combination or synthesis is not given “through objects” but can only be carried out by the subject (B130)’ (p. 159). But on the view sketched here, objects themselves do not instantiate the categories independently of their standing under the unity of apperception, and although this is not a result of the subject carrying out a process of combination, Kant does say that the combination of a manifold presupposes a unity which cannot arise from the combination but rather makes the combination possible (B130–1). Thus when Schulting says that ‘Kant’s own thesis [is] that the combination of objects is not a function of the objects themselves but exclusively of the subject of experience, and thus of judgement’ (p. 159), this reading is happy to accept everything except the last four words. The instantiation of the categories is not necessitated by the objects themselves, but by the fact that they stand under the unity of space and time, a unity which comes from ‘somewhere higher’ (B131) – namely, as this reading would have it, from the unity of apperception itself.

A similar response is available to the second worry. Schulting says that ‘if indeed the objects themselves necessitated their exemplification and it is not the subject that establishes the exemplification, how could I know that the categories are exemplified in an object?’ (p. 159). I think some of the issues here turn on aspects of Schulting’s own proposal, which I turn to below, and I am not sure that I get the full force of the objection. But someone who is sympathetic to the above reading will not accept that, on this reading, ‘it is not the subject that establishes the exemplification’; it is only that the subject does not establish the exemplification in virtue of a discursive act of categorial synthesis. The transcendental synthesis which accounts for the exemplification of the categories still has its seat in the understanding, and thus in the subject herself.

The third worry is that there is a structural problem for the proposal, because ‘if there is a fundamental problem with … the gap between (N1*) and (N3), then the “second step” cannot solve it’ (p. 160). Schulting’s thought here seems to be that the conclusion of the deduction – (N3) in his numbering, (O) in mine – is already established at §20 of the Deduction. He writes, ‘[i]f we look at Kant’s preliminary conclusion to the first half of the B-Deduction in §20 (B143), it is prima facie clear that Kant indeed claims to have shown that the argument for the necessity of [the transcendental unity of apperception] for any unitary manifold of representations implies that any such manifold is subject to the categories’ (p. 145). That is, ‘the fundamental synthetic connection with objects, which bridges the putative Gap, should already have been proved in the “first step”, with the “second step” only explicating how this synthetic a priori connection – the original-synthetic unity of apperception – has purchase in sensibility itself’ (p. 161).

I think the issues here turn on a question about how we should understand the second part of the B-Deduction. Schulting holds that the conclusion of the Deduction is already established by §20, and that the second part of the argument shows only how the categories are applied in spatiotemporal intuition. In contrast, the reading sketched above takes seriously Kant’s claims that the first and second parts of the B-Deduction are two parts of one proof, and instead holds that the objective validity of the categories is not established until §26. And if one wanted to put the opposing case, note that Kant tells us that ‘In the above proposition, therefore, the beginning of a deduction of the pure concepts of the understanding has been made . . . In the sequel (§26) … the aim of the deduction will first be fully attained’ (B144–5, my emphases). And one might worry that any reading which takes the second part of the Deduction to address the question of how the categories are applied in spatiotemporal intuition ends up reading into the Deduction the task of the Schematism and Principles.

4. The Gap Banned?

But, as I say, these issues are somewhat tangential, and the real substance of the chapter comes in Schulting’s own proposal. As I understand it, this has two aspects. First, Schulting argues that ‘Kant’s claim is that there is an inherent connection between [the transcendental unity of apperception] and [the objective unity of apperception], and hence between subjectivity and objectivity, or between the subjective conditions of experience (the functions of thought) and the objective conditions of experience (the categories)’ (p. 155). Second, he holds that ‘a suitably amended version of the phenomenalist reading of Kant’s transcendental idealism helps us understand the intimate relation between the unity of apperception and the concept of an object, and why there is no gap between the necessary application of the categories and their exemplification in the object of experience’ (p. 142).

These aspects are linked: Schulting holds that that it is the idealist thrust of the first Critique which explains why there is no gap to be bridged, and the failure to recognize this results from ‘a persistent general antipathy to idealism among Anglophone readers of Kant’ (p. 173). In very rough and ready terms, Schulting’s identification of the transcendental unity of apperception and the objective unity of apperception has it that there is no more to being an object – at least as considered formally and not materially – than being the outcome of an a priori synthesis which unites the manifold of sensations. This supports a moderate phenomenalist reading of Kant, one which holds that ‘the empirical object, that is, the object as appearance, only exists as construction out of representations, so exists only when and if it can be represented by a subject, any subject, in the realm of possible experience’ (p. 178).

There is much to say about this interesting proposal. But I want to relate it back to the question of whether the transcendental deduction contains a gap which needs to be bridged. Consider again the four responses to the distinction between (S) and (O) outlined above. Schulting endorses (3): that there is no gap to be bridged because in establishing (S) Kant ipso facto establishes (O). And his opponents endorse (4): that there is a gap to be bridged between (S) and (O).

Now on a natural way of approaching these two interpretative responses, the difference between (3) and (4) can seem enormous. Imagine calling in a builder to ask about a possible hole in your ceiling. There is a world of difference between the builder who tells you that there is a hole there, but that it can be fixed, and the builder who tells you that there isn’t really a hole there to fix. But there is an alternative perspective from which the divide between (3) and (4) can seem much less pressing. After all, we should all acknowledge that (S) is a distinct claim from (O) and that we need some argument to move from (S) to (O). Schulting takes it that the argument from (S) to (O) need not involve any additional claims, whereas his opponent thinks that the argument does require additional claims. Putting it somewhat tendentiously: Schulting thinks that the move from (S) to (O) is non-ampliative, requiring nothing more than is provided by a proper understanding of (S), whereas his opponent disagrees. So be it. But both think that an argument is needed to move from (S) to (O). Indeed, who could think otherwise?

It is in this minimal sense that the existence of a gap in the transcendental deduction seems to me non-negotiable, at least for those who take the third or fourth responses above. (In this sense, responses (3) and (4) have much more in common than either does with responses (1) and (2).) (S) makes a claim about our epistemic capacities; (O) makes a claim about objects themselves. Something must allow us to move from one to the other. The question is whether the gap is to be bridged by an argument which confines itself to a proper understanding of (S) or one which appeals to materials outside (S) itself.

This relates to the role of transcendental idealism in Kant’s argument. Schulting holds that a correct understanding of transcendental idealism will show that there is no gap to be bridged, that (S) entails (O) absent any other commitments. But his opponents also hold that transcendental idealism bridges the gap. The reading sketched above, for instance, takes the gap to be bridged in virtue of a dependence of the unity of space and time on the unity of apperception. This is a form of idealism, perhaps one which is more metaphysically loaded than the moderate phenomenalism which Schulting endorses, but a form of idealism nevertheless. Thus, whilst I happily acknowledge a persistent general antipathy to idealism (see e.g. Gomes Reference Gomes2017a), I demur from the suggestion that the proposed solution to the gap in my (2010) makes no appeal to idealism: both Schulting’s proposal and the alternative take the move from (S) to (O) to be bridged by idealism. Indeed, who could think otherwise?

This brings us back to the debate between Stroud and Strawson since one of Stroud’s points against Strawson is that the gap he identifies can be bridged only by some form of idealism or verificationism, neither of which Strawson wishes to adopt. It seems to me that both Schulting and his opponents recognize that transcendental idealism is the key to any move from (S) to (O). The question is what form such idealism takes, and how it enables Kant to bridge that gap. Schulting’s chapter, and his book as a whole, contain a stimulating answer to that question. It is on this territory, rather than on the question of whether a gap exists in the deduction, that competing accounts will be assessed.Footnote ¹