In section 1, I explain the need for an account of what Kant means by ‘cognition’ (Erkenntnis). In section 2, I use Kant’s texts and those of his German predecessors to show that he understands empirical cognition to be rational sensory discrimination. In section 3, I provide evidence – partly from the eighteenth-century German translation of Hume’s Enquiry – that Kant read and understood Hume as granting the assumption that we have empirical cognition. In section 4, I turn to the text of the A- and B-editions of the Transcendental Deduction and show that Kant’s fundamental assumption is that we have empirical cognition.
1. The Problem: What is Kant's Starting Point?
The Transcendental Deduction of the first Critique, at least in the A-edition, consists of two intertwined strands of argument (Axvi–xvii).Footnote 1 The first strand is the objective deduction, which aims to show that the categories – a priori concepts including ‘substance’, ‘cause’, and ‘effect’ – have ‘objective validity’. This would show, in effect, that we are justified in applying them to objects of experience. The second strand is the subjective deduction, the purpose of which has been a matter of scholarly debate. I focus here on the objective deduction, which Kant takes to achieve the essential task of the transcendental deduction. Complicating matters is that Kant rewrote nearly entirely the Transcendental Deduction chapter for the B-edition. Nevertheless, my paper generally focuses on what is common to both editions.
Both editions of the Transcendental Deduction share a nearly identical first section, On the Principles of a Transcendental Deduction in General (A84–94/B117–29). This section presents an explanation of the need for a transcendental deduction (A84–92/B117–24) and contains a subsection, Transition to the Transcendental Deduction of the Categories (A92–4/B124–9). In the Transition, Kant clarifies as follows his argumentative strategy for showing the categories have objective validity.
The transcendental deduction of all a priori concepts therefore has a principle toward which the entire investigation must be directed, namely this: that they must be recognized as a priori conditions of the possibility of experiences (whether of the intuition that is encountered in them, or of the thinking). (A94/B126, my emphasis)
According to the principle Kant presents here, the transcendental deduction’s task of demonstrating the categories’ objective validity can be accomplished if those concepts are shown to be a priori conditions for the possibility of experiences. He provides two routes by which a priori concepts can be shown to be a priori conditions of the possibility of experiences: to show they are a priori conditions for either the intuition or the thinking encountered in experiences. Neither of these routes pertains to intuition or thinking generally but to intuition or thinking encountered in possible experiences, i.e. to either the intuitive aspect or the conceptual aspect of experience.
As stated at A87/B119–20, the concepts of space and time were given a transcendental deduction. This refers to the Transcendental Aesthetic, which can be understood as taking the first route suggested in the above passage: it begins with the intuitive aspect of our experience (its spatiotemporal character) and shows this is possible only if space and time are a priori forms of our intuition. In contrast, the transcendental deduction of the categories takes the second route mentioned in the above passage.Footnote 2 It demonstrates the categories’ objective validity by showing they are a priori conditions for the possibility of the thinking involved in our experiences.
It is possible to interpret the transcendental deduction in another way: namely, that it demonstrates the categories’ objective validity by showing they are a priori conditions for both intuition and thought encountered in experience. This might be seen in Kant’s beginning with ‘unity of intuition’ in the A-edition (A99) and in his footnote in the B-edition suggesting that the unity of intuition is made possible by the understanding or the faculty of thought (B160–1). But even if the categories are shown to be a priori conditions for the unity of intuition, the transcendental deduction of the categories should be taken as following the second route described in the passage at A94/B126. This is because Kant begins each edition’s transcendental deduction by arguing that the basic unity had by intuition is conceptual insofar as it is made possible by the understanding. From A98–100, Kant argues that intuition’s unity is made possible by the synthesis of apprehension, an activity of the understanding. Similarly, from B129–31 he argues that all combination of the manifold of intuition, and with it any unity it has, cannot come from intuition alone but requires the understanding. These passages suggest that Kant takes the second route insofar as he first argues that the unity of intuition encountered in experience is made possible by the understanding and then goes on to argue that this activity of the understanding requires the categories. On this basis, then, Kant holds that the categories apply universally to all intuition encountered in experience.Footnote 3
If the transcendental deduction of the categories takes the second route (or alternatively quickly argues that the understanding is responsible for the unity of intuition), then it would assume our experience involves thought. But what exactly is the type of experience assumed by Kant’s principle for the transcendental deduction? He ultimately claims all experience consists in the empirical cognition of objects (B147, Anth., 7: 141, R5661, MM, 29: 804, 816).Footnote 4 As we shall see later, it is a matter of debate whether he argues that our experience is a cognition of objects or whether he assumes that it is such.Footnote 5 Surprisingly, one of the essential tasks for settling this issue has been rarely carried out until recently, namely to work out an explicit account of what Kant understands by ‘cognition’.Footnote 6 Although most commentators hold that empirical cognition is an epistemic state that requires both intuition and concepts, few sufficiently explain the sense in which empirical cognition counts as an epistemic success. For example, in her recent interpretation of the transcendental deduction, Patricia Kitcher gives persuasive textual evidence that Kant holds cognition to require concepts and the capacity for being aware of reasons (Reference Kitcher2011: 118–21). Likewise, Henry E. Allison states that (discursive) ‘cognition consists in the rule-governed act of taking x as F’ (Reference Allison2015: 284).Footnote 7 Although both of these statements are correct, neither elucidates what I take to be fundamental about Kant’s notion of cognition, namely the manner in which cognition counts as an epistemic success. After providing an interpretation of cognition along these lines, I argue that the transcendental deduction starts with the assumption that our experience consists in empirical cognition so understood.
2. Kant’s Conception of Empirical Cognition
I provide my account of empirical cognition in five steps. I first show in section 2.1 that Kant uses the notion of cognition in a broad sense to refer to any mental state that refers to an object. Section 2.2 shows that there is a conceptual link between a stricter sense of empirical cognition – one which is more central to Kant’s epistemological concerns – and the notions of clarity and distinctness. I then present in section 2.3 the basic Leibnizian-Wolffian conception of clarity and distinctness. In section 2.4 I use the preceding distinctions to present Kant’s stricter conception of empirical cognition. In section 2.5 I explain that this notion of empirical cognition expresses an epistemological paradigm of ‘rational sensory discrimination’.
2.1 Kant’s Broad Sense of Cognition
In the first Critique, Kant presents a Stufenleiter – or progression of types of representation – that exhibits what I call his broad sense of cognition (A319–20/B376–7). The Stufenleiter treats representation as a genus with two species: those accompanied with consciousness and those that are not. The former are perceptions, which are in turn classified as subjective – sensations referring to the state of the subject – or objective – cognitions referring to the object. Hence, according to the Stufenleiter (as well as many other passages), a cognition is an objective perception, i.e. a conscious representation relating to an object (Bix–x, xvii–xviii; R1685, 1693, 2128, 2836, 3055, 5221). Kant perhaps inherited this broad notion of cognition from Meier’s logic textbook, which associates cognition with the representation of an object (1752: §§10–11).Footnote 8
Unfortunately, this characterization of cognition is not particularly helpful for understanding Kant’s views in first Critique. For the Stufenleiter classifies intuitions and concepts each on their own as counting as cognitions. This at least seems to clash with the doctrine of the first Critique that neither intuition alone nor concepts alone are sufficient for cognition (A51–2/B75–6). But in a late work, Kant explains his peculiar use of terminology in the Stufenleiter: even though both intuition and concepts are required for cognition, a cognition can be described as either an intuition or a concept ‘after that to which I particularly attend on each occasion, as the determining ground [of the cognition]’ (WF, 20: 325).Footnote 9 Thus although we have learned from the Stufenleiter that a defining characteristic of cognition is that it is a representation relating to an object, that passage fails to shed light on what is distinctive about cognition and exactly how it requires both intuition and concepts. But Kant uses many terms, for example, ‘understanding’, in broad and stricter senses. I next provide two passages that offer clues about a stricter sense of cognition.Footnote 10
2.2 Kant’s Stricter Sense of Cognition
First, Reflexion 2394 gives the following degrees of representation.Footnote 11
The following degrees are to be distinguished:
1. to represent something.
2. to know something. To represent with consciousness. (Later addition: representing to oneself with consciousness. …
3. to be acquainted with something. Thereby in comparison to differentiate from others.
4. to understand something. (Later addition: what I am acquainted with and understand, that I cognize. (Later addition: to be able to expound and communicate to others.) to cognize something through the understanding. … (R2394, my translation)
Kant introduces the notion of cognition in his presentation of the fourth degree: to understand something. There he claims that acquaintance and understanding are jointly sufficient conditions for cognizing something: ‘what I am acquainted with and understand, that I cognize’. This claim gestures at Kant’s doctrine that cognition requires both intuition and concept. For it seems that while intuition would be required for us to be acquainted with things so as to differentiate them, concepts would be required for us to understand those things.Footnote 12
Second, Kant’s Anthropology relates the notions of clarity and distinctness to cognition:
Consciousness of one's representations that suffices for the differentiation of one object from another is clarity. But that consciousness by means of which the composition of representations also becomes clear is called distinctness. Distinctness alone makes it possible that an aggregate of representations becomes cognition, in which order is thought in this manifold, because such a composition with consciousness presupposes unity of that consciousness, and consequently presupposes a rule for that composition. … in every cognition (since intuition and concept are always required for it), distinctness rests on the order according to which the partial representations are combined … (7: 137–8, my translation; cf. R2281)
This passage affirms the first Critique’s doctrine that both intuition and concept are required for cognition, so it is a good source for understanding the stricter sense of cognition (cf. Anth., 7: 140). Yet whereas R2394 puts forward acquaintance and understanding as jointly sufficient for cognition, this passage states that clarity and distinctness give rise to cognition. As I shall argue, R2394’s treatment of cognition in terms of acquaintance and understanding can in turn be explicated with the Anthropology’s notions of clarity and distinctness. My main interpretative thesis is then that empirical cognition in the stricter sense is a clear and discursively distinct representation, where clarity is provided by means of acquaintance and distinctness by means of understanding.
Unfortunately, the first Critique does not have much to say about these notions or how they relate to cognition. Kant’s characterizations of cognition in that key Critical-period work are various and often present the notion in very general terms. For example, we find the following two characterizations in each edition’s Transcendental Deduction: ‘a whole of compared and connected representations’ (A97) and ‘a determinate relation of given representations to an object’ (B137).Footnote 13 These characterizations prompt many questions: what sort of connections obtain between representations in a cognition, and what sort of determinate relation must representations have to objects in a cognition? Of course, part of Kant’s task in the first Critique is to offer detailed answers to these questions. But many of these answers are arrived at through contentious argumentation. My approach to this problem – namely that Kant’s characterizations of cognition are either too vague or too loaded – is to try to understand what he would have taken his potential interlocutors as willing to grant about the nature of our cognitive engagement with the world. Kant’s target audience included thinkers with pre-Critical views and terminology, and for him to persuade that audience of the Critical philosophy, he must assume at least some of those views and use some of that terminology. So I proceed by investigating how pre-Critical German philosophical thought, including that of Kant himself, understood some of the key notions involved in characterizing our experience. Accordingly, I begin by outlining the German philosophical heritage behind the notions of clarity and distinctness.Footnote 14
2.3 The Leibnizian-Wolffian Conception of Clarity and Distinctness
Obviously, Descartes marks the starting point in early modern philosophical use of the notions of clarity and distinctness to describe certain sorts of epistemic success. But Kant’s understanding of these notions was filtered through the German philosophical tradition. This began with Leibniz, whose first published philosophical essay – ‘Meditations on Knowledge (cognitio), Truth, and Ideas’ – offers a different account from Descartes of the notions of clarity and distinctness (Leibniz 1989). Leibniz was satisfied with his characterizations of these notions much later in his career, as attested by his presentation of them in New Essays Concerning Human Understanding, a work which Kant read soon after its posthumous publication (Leibniz Reference Leibniz1996: 254–6).Footnote 15 It is clear that Leibniz’s distinctions were already influential before the publication of the New Essays, as attested in the works of Baumgarten, Meier, Lambert and Eberhard.Footnote 16 But both Kant and these thinkers were likely more familiar with Christian Wolff’s presentation of the two notions than with Leibniz’s, as Wolff’s works were much more accessible and widely read.Footnote 17 Accordingly, I limit my focus to Wolff’s account of clarity and distinctness as presented in his German Metaphysics, with which Kant was quite familiar.
Before turning to the notions of clarity and distinctness themselves, we ought to consider what Wolff applied them to. In the German Metaphysics, Wolff applies the notions of clarity and distinctness to thoughts: conscious representations of objects (Wolff Reference Wolff2003: §§198–214).Footnote 18 Recall that this focus on how objects are represented is distinctive of what I argued to be Kant’s broad conception of cognition. Although Wolff himself is concerned with thoughts, I shall present his conceptions of clarity and distinctness with respect to the more generic ‘representation’. This is to set aside some terminological differences between Wolff and Kant, as well to avoid tricky issues concerning eighteenth-century views of animals’ representational capacities.
Let us first consider Wolff’s characterization of clarity: ‘Some thoughts are so constituted, that we quite well know what we think and can distinguish them [what we think] from others. In that case let us say that they are clear’ (§198, my translation). Clarity for Wolff first requires that one is conscious of the objects of one’s representations, since the surrounding passages suggest that to ‘know what we think’ is equivalent here to being conscious of what we think. Clarity also requires the ability to distinguish objects from one another. For example, my parents’ dog Duke represents me clearly insofar as he is conscious of me through the representation and can thereby differentiate me from a stranger (as indicated by his growling at the stranger but not at me). We can characterize clarity, then, as follows:
A subject’s representation of something is clear just in case that subject, by means of that representation,
(i) is conscious of the object; and,
(ii) can differentiate it from other things.
Let us now turn to distinctness. Wolff writes that it requires clarity plus something more: ‘When it happens that we determine the difference of that which we think and can thus also on demand say it to others. And thereupon are our thoughts distinct’ (§206, my translation). Wolff explains that distinctness requires clarity because the former arises out of the clear representation of a thing’s parts (§207). His examples suggest that a subject’s clearly representing a thing’s parts amounts to her ‘determining the difference of what she thinks’ insofar as she is able to cite the similarity or difference of a thing’s parts as grounds or reasons for why the thing she distinctly represents is the same as or different from other things (§207). For example, unlike Duke, my friend Ingrid represents me not only clearly but also distinctly. Her representation of me is clear because she is conscious of me through it and differentiates me from other things. It is also distinct because she can determine wherein I differ from other things. Insofar as she is aware of my parts – e.g. my having glasses, brown hair, etc. – she can become conscious of the reasons for her determination of me as the same as or different from other things. And as the above passage suggests, Ingrid’s capacity for distinctly representing me is bound up with her capacity to make judgements to communicate the similarities and differences between me and other things. Accordingly, we can define distinctness as follows:
A subject’s representation of something is distinct just in case that subject, by means of that representation,
(i) is conscious of the object;
(ii) can differentiate it from other things;
(iii) on the basis of a clear representation of the thing’s parts, can determine wherein that thing differs from other things; and,
(iv) can express judgements giving reasons for such differentiations.
Wolff explicitly grounds the capacity for distinct representation on the faculty of understanding, which is a higher cognitive faculty raising us above mere animals and which relates to our conceptual and judgemental capacities. He further claims that the faculties of the senses and imagination are distinguished from the understanding insofar as the representations of the former two faculties can never become distinct without the understanding (§§277ff.).
2.4 Kant's Conception of Empirical Cognition in Terms of Clarity and Distinctness
I now argue that in Kant’s stricter sense, empirical cognition is a clear and discursively distinct representation, where those notions are to be understood in roughly Leibnizian-Wolffian terms. I do so by showing how acquaintance and understanding are respectively to be understood in terms of clarity and distinctness.
Let us first consider acquaintance and clarity. There are two reasons to identify acquaintance with clear representation. First, several of Kant’s characterizations of acquaintance fit the Leibnizian-Wolffian model of clarity, which requires that one be aware of something by differentiating it from other things.Footnote 19 For example, in R2394 Kant writes that when one is acquainted with something, one is ‘thereby to differentiate [that thing] in comparison from others’.Footnote 20 Second, this conception of acquaintance fits Kant’s own characterizations of clarity across many works. For example, the Logik Blomberg and the Metaphysik Mrongovius both discuss what it means to represent the Milky Way and its parts (its constitutive stars) clearly and distinctly (LB, 24: 41, 119; MM, 29: 879–80). When we fail to represent the stars of the Milky Way clearly – i.e. when we represent them ‘obscurely’ – we fail to differentiate the stars from one another. Hence, a failure to represent something clearly amounts to a failure to be acquainted with that thing. Accordingly, there is good reason to say acquaintance and clear representation are equivalent.Footnote 21
Let us now consider understanding and distinctness. There is textual support showing that distinctness is made possible by the understanding. Recall that Wolff explicitly grounds the capacity for distinct representations on the faculty of understanding. As exhibited in the Anthropology passage cited earlier, Kant too views the distinctness required for cognition as arising by means of three things: ‘thought’, ‘unity of consciousness’ and ‘a rule for composition’. The understanding, however, is responsible for all three: thought in the form of judgement (A69/B94), unity of consciousness in the form of apperception (A119) and rules in the form of concepts (A132/171). Furthermore, in a pre-Critical essay, Kant claims that the capacity to judge, i.e. the understanding, is what distinguishes us from mere animals, and his explanation appeals to the view that the understanding provides us with distinct representations (FS, 2: 59–60). He claims that it is possible for non-human animals to differentiate things ‘physically’ by having something like a clear representation of the characteristic marks of things. Presumably, ‘to differentiate physically’ is to act through ‘instinct’, which Kant elsewhere describes as the principle of animal life and is ‘the faculty for performing actions without consciousness’ (MD, 28: 689–90). Hence, for animals ‘to differentiate physically’ is for them to act out of non-conscious physical impulse in response to clear representations of the characteristic marks of things. We, in contrast, are also able to differentiate things ‘logically’ with distinct representations. Kant claims that a distinct representation requires that a characteristic mark of a thing be ‘cognized as a characteristic mark of the thing’ (FS, 2: 58, my emphasis). So unlike animals, we can be aware of why characteristic marks differentiate things. We can do this because we have the capacity to judge (FS, 2: 59). In other words, the understanding provides us with distinct representations insofar as we use the understanding to make judgements using concepts.
We have good reason, then, to interpret Kant as identifying understanding something with distinctly representing something. But there is an important qualification. Kant mentions in several places that there is both intuitive or sensible distinctness as well as conceptual or discursive distinctness (Anth., 7: 135; LB, 24: 42; R1690, 1709, 1821, 2363). I do not have space to address what intuitive or sensible distinctness is or why Kant introduces it. But the evidence above indicates that understanding requires discursive distinctness, i.e. a distinctness arising from the use of concepts in judgements.
Let us now use the preceding to shed light on the notion of empirical cognition. Recall Kant’s claim from R2394 that ‘what I am acquainted with and understand, that I cognize’. From this we know that acquaintance and understanding are jointly sufficient for cognition. If we take acquaintance and understanding to give rise respectively to clarity and distinctness, a passage cited earlier from the Anthropology suggests that acquaintance and understanding are also jointly necessary for empirical cognition: ‘[d]istinctness alone makes it possible that an aggregate of representations [which aggregate is clearly represented] becomes cognition’ (7: 137–8, my italics). If my account is correct, then a subject has empirical cognition just in case she has a representation that is both clear and discursively distinct – i.e. distinct by means of logical or conceptual differentiation. Formulated with less jargon, this means that a subject empirically cognizes an object just in case she identifies that object or differentiates it from other things (whether correctly or not) through the formation of judgements (whether correct or not).
2.5 Kant’s Epistemological Paradigm
This account shows the most fundamental sense in which empirical cognition counts as an epistemic success. Namely, it can be described as ‘rational sensory discrimination’: we have empirical cognition insofar as we discriminate between objects given through the senses and insofar as those discriminations are based on judgements (and hence sensitive to reasons). The significance of this is that it makes explicit Kant’s epistemological paradigm. For example, we can understand better why his paradigm is not knowledge in the sense of justified true belief. This is important for three reasons. First, many Kant scholars still misleadingly use the terms ‘knowledge’ (Wissen) and ‘cognition’ (Erkenntnis) interchangeably.Footnote 22 Doing this is especially problematic if they are to make themselves clear to non-specialists. Second, although some have explicitly asserted that cognition is not justified true belief or noted that Kant’s Wissen is more akin to the notion of knowledge,Footnote 23 hardly any provide a detailed account of what cognition consists in. Third, an explicit and accurate account of cognition can help us avoid bringing false assumptions to bear on Kant.Footnote 24
Recognizing that empirical cognition amounts to rational sensory discrimination also helps us see how Kant assumes in the Transcendental Deduction that our experience consists in empirical cognition. Without a detailed account of empirical cognition, it is easy to treat it as a type of knowledge that a sceptic could doubt we have; it would thus seem to be an improper starting point for the Transcendental Deduction.Footnote 25 But in the next section I show how Kant would have taken an assumption of empirical cognition to be particularly well-suited for his attempt in the first Critique to respond to a form of Humean scepticism with which he was concerned.
3. Cognition and the Humean Problem
In this section, we are not concerned with what Hume actually thought but rather with how Kant understood Hume, whether correctly or not, and with how he might have taken himself to have responded to him. In the Prolegomena, Kant claims the first Critique is ‘the elaboration of the Humean problem in its greatest possible amplification’ (4: 261). Ultimately, Kant saw Hume as challenging not merely our knowledge of causality but the possibility of synthetic a priori cognition in general (4: 277). Accordingly, we can call ‘the Humean Problem’ the question whether we have cognition of synthetic a priori principles that ground our experience. Synthetic a priori cognition expresses the transcendental conditions necessary for the possibility of experience.Footnote 26 Given my interpretation, this means that synthetic a priori cognition expresses the conditions that make possible rational sensory discrimination.
Since the Transcendental Deduction is an essential part of Kant’s answer to the question of the possibility of synthetic a priori cognition, it seems that he ought to have used premises he would have taken Hume to concede. And there are indeed reasons to think that Kant would have interpreted Hume as conceding that we have empirical cognition. It is known that Kant was more familiar with Hume’s Enquiry than the Treatise, although he had some acquaintance with the latter via Beattie, Hamann and Tetens.Footnote 27 Nevertheless, little emphasis has been given in the secondary literature to the German translation of Hume’s Enquiry that Kant read. The translation was edited by Johann Georg Sulzer and was based on the second edition of Hume’s work.Footnote 28 The second edition’s title was Philosophical Essays Concerning Human Understanding, and the translation of this title is noteworthy: Philosophische Versuche über die Menschliche Erkenntniß. The translated title replaces understanding (Verstand) with cognition (Erkenntnis), and this suggests that the Hume read by Kant was expressly concerned with cognition.
We have, moreover, reasons to think Kant read Hume as granting that we have empirical cognition. In Essay 4 of the Enquiry, Hume presents his sceptical challenge about causality. A key claim Kant would have read in the German translation is the following:
I will venture, as a universal proposition which admits no exceptions, to claim, that this cognition of cause and effect (diese Erkenntniß der Ursache und Wirkung) is achieved in not one example through rational inferences a priori; rather it comes entirely from experience, in virtue of which we find that particular objects are constantly united (vereinigt) with another. (1755: 68, my translation back into English, as well as my emphasis)Footnote 29
Kant was often not a close reader of other philosophers’ texts, and I suggest that passages such as the above could have been read by Kant as admissions by Hume that we have empirical cognition of cause and effect. Perhaps Kant might not have understood Hume’s use of Erkenntnis as equivalent to the strict sense of empirical cognition presented above. But in whatever way Kant understood Hume to use the term in question, passages such as the above match his understanding of Hume’s specific challenge about cognition. The Prolegomena states that Hume did not question whether the concept of cause is right, useful or indispensable for cognition of nature (4: 258–9). Rather, the challenge is to determine the sources of our cognition in general, including that of cause and effect. The above passage suggests that this is Hume’s concern, since in it Hume denies that rational inferences a priori are the source of our cognition of cause and effect. Later, Hume extends this by claiming that the source of our experience of cause and effect is also not the understanding (1755: 78). Kant’s Humean Problem, then, is that our empirical cognition – particularly of cause and effect, but also more generally – does not have a priori sources in the understanding or reason.
Kant’s assumption that we have empirical cognition is well-suited for responding to this challenge. This is because he likely understood Hume as granting each of the two components of my definition of cognition. Recall first that cognition requires that a subject discriminate between objects. Throughout Hume’s presentation of his sceptical doubts, he admits that our experience is of objects (Dinge or Gegenstände). Of course, in using the term ‘objects’, Hume is not committed to the existence of anything mind-independent. Likewise, although for Kant empirical cognition is of objects, this does not mean that those objects are mind-independent. Hume also grants that objects of experience are connected and combined in time through laws of association. This would seem to require that such objects are identified and differentiated from one another, as occurs with Kant’s empirical cognition. This can be seen by considering Hume’s treatment of causality in terms of constant conjunction. Hume admits that we do find particular objects to be constantly conjoined with others. For us to do this, we must recognize that the first set of objects are the same in some way, that the second set of objects are likewise the same in some way, and finally that the two sets of objects are different in some way. Thus Hume’s analysis of causality requires that he grant us the ability to identify and differentiate objects, i.e. to be acquainted with objects or to represent them clearly.
Hume also seems to be committed to holding that we can become aware of the judgements on the basis of which we treat objects as the same or different. This is a pretty minimal requirement. For it only requires that a subject be capable of judging why things seem to her to be the same or different, and it does not matter whether or not those judgements are correct, i.e. it does not matter whether a subject is justified. It would be implausibly extreme for Hume to deny that we have such capacities, and in fact he himself notes that we are capable of giving reasons for believing matters of fact (1755: 67–8). Hence, Hume also can be read as granting us the ability to understand objects or to represent them distinctly.
We have good reason, then, to think Kant would have taken Hume to grant that we have empirical cognition. If this is correct, then we should take more seriously the possibility that the Transcendental Deduction is capable of responding to sceptical challenges, a reading that has been repeatedly challenged in Kant scholarship.Footnote 30 My interpretation of empirical cognition and examination of Kant’s understanding of Hume provides evidence that the argument can respond to at least one particular sceptical challenge, ‘the Humean problem’.
4. Interpreting Kant’s Starting Points
Many prominent commentators have taken Kant’s fundamental starting point in the Transcendental Deduction to be consciousnessFootnote 31 or some form of self-consciousness, from which starting point he later derives that we have empirical cognition.Footnote 32 In this section, I show that the text of both the A- and B-editions of the first Critique more plausibly supports interpreting Kant as starting with the more basic assumption that we have empirical cognition.Footnote 33 Although Kant conceives of empirical cognition as having both consciousness and a form of self-consciousness (namely, transcendental apperception) as necessary conditions, his primary arguments in the Transcendental Deduction begin with the assumption that we have empirical cognition.Footnote 34
4.1 A-Edition Evidence
In the A-edition, Kant presents what seem to be two versions of the objective deduction, the so-called ‘argument from above’ and ‘argument from below’. Kant begins the argument from above as follows:
Now if we wish to follow the inner ground of this connection of representations up to that point in which they must all come together in order first to obtain unity of cognition for a possible experience, then we must begin with pure apperception. All intuitions are nothing for us (für uns nichts) and do not in the least concern us (gehen uns nicht im mindesten etwas an) if they cannot be taken up into consciousness, whether they influence it directly or indirectly, and through this alone is cognition possible. (A116, my italics)
On a cursory reading, it might seem as if this passage supports interpreting Kant as starting from the assumption that we have self-consciousness, namely pure apperception. But that semblance is mistaken, for Kant claims ‘we must begin with pure apperception’ if we want to know what is necessary for the ‘unity of cognition for a possible experience’. Kant claims that pure apperception is a necessary condition for cognition, and this suggests that he assumes we have cognition. This is reinforced by how he continues: intuitions must be ‘taken up into consciousness’ for cognition to be possible; otherwise, they ‘are nothing for us’. There is evidence that representations being ‘nothing for us’ means they fail to count as cognition or objects of cognition.Footnote 35 This is most explicit in a letter from 1789 to Markus Herz: ‘if intuitions (of objects of appearance) did not agree with these conditions, those objects would be nothing for us (für uns nichts), that is, not objects of cognition at all’ (11: 51, my emphasis).
The ending of the argument from above also indicates that Kant’s starting assumption is that we have cognition:
Now since this relation of appearances to possible experience is likewise necessary (since without it we could not obtain any cognition at all through them, and they would thus not concern us at all (sie uns mithin gar nichts angingen)), it follows that the pure understanding, by means of the categories, is a formal and synthetic principle of all experiences, and that appearances have a necessary relation to the understanding. (A119)
Here, Kant parenthetically remarks that the relation of appearances to possible experience is necessary because otherwise neither would we have cognition nor would appearances concern us (presumably, Kant does not draw a distinction between not concerning us and being nothing for us). So again, the text indicates that Kant’s basic assumption is that we have empirical cognition.
Kant begins his so-called ‘argument from below’ with the following:
Now we will set the necessary connection of the understanding with the appearances by means of the categories before our eyes by beginning from beneath, namely with what is empirical. The first thing that is given to us is appearance, which, if it is combined with consciousness, is called perception (without the relation to an at least possible consciousness appearance could never become an object of cognition for us (würde Erscheinung für uns niemals ein Gegenstand der Erkenntnis werden können), and would therefore be nothing for us (für uns nichts), and since it has no objective reality in itself and exists only in cognition it would be nothing at all). (A119–20, my emphasis)
The argument from below seems to begin with the assumption that our experience consists in appearances, as Kant claims ‘[t]he first thing that is given to us is appearance’. Kant then notes that he is concerned with appearances insofar as they are combined with consciousness, i.e. with what he calls ‘perceptions’. Hence, this might further seem to support interpreting Kant as beginning with the assumption that we have consciousness. These semblances are misleading. Kant begins the argument from below with appearances or perceptions because his argumentative strategy is to assume we have empirical cognition, which itself requires perceptions.Footnote 36
There are several reasons for preferring this interpretation. First, Kant’s parenthetical remark explains why perceptions – in the sense of appearances combined with consciousness – are worthy of being examined for his present purposes. An appearance without consciousness – i.e. one that is not a perception – would be ‘nothing for us’ because it ‘could never become an object of cognition for us’. For this reason, appearances without consciousness are not of present interest. One might still take this passage as beginning with the assumption that we have perception. But if that were so, I can see no reason why Kant would have included this parenthetical remark. More importantly, better sense is made of how the arguments from above and below cohere if we interpret Kant’s primary assumption as being that we have empirical cognition, rather than the assumption that we have consciousness or perceptions. For we already saw evidence that Kant begins the argument from above by assuming we have empirical cognition. And in beginning each of the arguments, he sets aside what is ‘nothing for us’ to consider what is relevant for empirical cognition. Finally, with my interpretation of empirical cognition we can see why Kant is able to offer both an argument from above and an argument from below. For on this interpretation, empirical cognition requires both a higher cognitive faculty for discriminating between objects (the understanding) and material given by a lower cognitive faculty (perceptions). This makes it possible for Kant to assume that we have empirical cognition and then to carry out on the basis of this assumption two arguments: one investigating the conditions for the higher cognitive faculty (the argument from above) and the other investigating the conditions for being given material from which that higher cognitive faculty can make discriminations of objects (the argument from below).
If one were to interpret the primary assumption of the argument from below as being that we have perceptions, then it is hard to see how that argument is supposed to cohere with the argument from above. But if we take the assumption of perception as based on the deeper assumption that we have empirical cognition, then a more unified account of the two arguments is possible. The assumption of the argument from below – that we have consciousness or perceptions – would then be explained by Kant’s concern with the necessary conditions of cognition, one of which is consciousness.
So far, I have shown that the arguments from above and below are more plausibly read as beginning with an assumption of empirical cognition, rather than an assumption of self-consciousness or consciousness. One could object that even if these passages show that Kant relies on the thesis that we have empirical cognition, he nevertheless arrives at this thesis by means of other more fundamental assumptions and argumentation. Yet I cannot find any place in either the argument from above or from below where Kant argues to the conclusion that we have empirical cognition based on the alternative assumptions of self-consciousness or consciousness.
Consider the possibility that in the A-edition Kant offers an argument for empirical cognition based on the assumption that we have consciousness. The best place to find such an argument would likely be the earlier subjective deduction. In fact, Robert Paul Wolff, who advances the interpretation that Kant’s fundamental assumption is that we have consciousness, finds that the subjective deduction is essential to understanding the Transcendental Deduction’s argument for the categories’ objective validity (Wolff Reference Wolff1963: 80). But the subjective deduction does not begin with the assumption of consciousness. Kant’s first mention of ‘consciousness’ comes only in the third stage of the subjective deduction (A103), and this indicates that Kant did not assume in the earlier stages anything about consciousness. Furthermore, Kant introduces the subjective deduction as an investigation preparing the reader for the Transcendental Deduction’s ‘deep penetration into the primary grounds of the possibility of our cognition in general’ (A98, my emphasis). Thus Kant’s primary concern in the subjective deduction is cognition and not consciousness.
Consider then the possibility that the A-edition offers an argument for empirical cognition based on the assumption that we have some form of self-consciousness. This is implausible because Kant first introduces a notion of self-consciousness in his argument that we have transcendental apperception, since it is a necessary condition for empirical cognition (A106–7). Hence, the assumption of cognition is more fundamental than one of self-consciousness.
4.2 B-Edition Evidence
The following passage is often read as evidence that Kant’s starting point in the transcendental deduction is transcendental apperception:Footnote 37
The I think must be able to accompany all my representations; for otherwise something would be represented in me that could not be thought at all, which is as much as to say that the representation would either be impossible or else at least would be nothing for me (für mich nichts). (B131–2, my italics)
It is often overlooked that Kant’s first proposition here – that the ‘I think’ must be able to accompany all my representations – is supported by an argument. Kant’s argument is that this proposition must be true or else some of my representations ‘could not be thought at all’, and such representations would ‘be impossible’ or ‘at least would be nothing for me’. Although this argument blatantly seems to be about the nature of representations in general, or at least ‘my’ representations, I shall argue that it is better made sense of by interpreting Kant as assuming both that we have cognition and that cognition requires thought.
Consider first the following claim from the above passage: if it is not necessary that the ‘I think’ be able to accompany all my representations, then ‘something would be represented in me that could not be thought at all’. This means that Kant treats transcendental apperception as a necessary condition for thought,Footnote 38 where ‘thought’ is to be understood for Kant – as evidenced by the earlier metaphysical deduction (A69/B94) – as a mental activity that applies concepts in judgements.
Consider now Kant’s next claim: the representing of something in me without thought is equivalent to that representation being either ‘impossible’ or ‘nothing for me’. If we read this passage in isolation, then it seems that Kant refers here to either representations in general or representations in me. Neither interpretation is plausible. First, if we consider representations in general, then we must note that Kant himself describes animals as having representations but neither transcendental apperception nor thought.Footnote 39 Hence, Kant holds that it is not only possible but actual that there are representations without thought. Of course such representations without thought might in some sense be ‘nothing for me’, but this would not ground an argument for the claim that it is necessary for ‘I think’ to be able to accompany all my representations. Second, consider representations in me. Kant admits that human beings have unconscious representations, so this makes it likely that he holds it to be possible that a representation is in me without being thought.Footnote 40 Perhaps then Kant means to talk about only those representations in me that are not ‘nothing for me’. But what exactly does that mean? As I have already pointed out, there is evidence that Kant describes representations as ‘nothing for me’ when those representations are not part of cognition. This suggests, then, that the above passage ought to be taken as pertaining to cognition.
Accordingly, let me now show how the passage can be made understandable by interpreting Kant as having in mind empirical cognition. Recall that Kant’s argument runs as follows. If it weren’t the case that the ‘I think’ must be able to accompany my representations, then there would be a representation in me that could not be thought. But such a representation – one in me that could not be thought – is either impossible or nothing for me. On my interpretation, we can make sense of why Kant offers this disjunction. Namely, consider whether the representation in question counts as cognition or not. First, suppose this representation counts as cognition. Such a representation is impossible, as my account of empirical cognition shows. For insofar as empirical cognition requires discursive distinctness, it requires thought, i.e. judgements that apply concepts. Second, suppose the representation in question does not count as cognition. As we have seen, Kant describes representations as ‘nothing for me’ to indicate that a representation does not contribute to cognition. And accordingly, Kant states the same about the representation in question: if it does not count as cognition, then it is nothing for me.
As we have just seen, interpreting Kant as assuming we have cognition makes sense of his claim that a representation that is not thought is either impossible or nothing for me. On this interpretation, the opening of §16 aims to show that cognition has as its necessary condition transcendental apperception. Kant shows this by relying on the theses that cognition requires thought and that thought requires transcendental apperception.
Given this reading of the opening of §16, we can understand better why it is preceded by the remarks of §15, in which Kant discusses ‘the possibility of a combination in general’. Kant discusses this because he is assuming that all our experience involves the combination of a manifold, and it is natural for him to make this assumption if he understands experience in terms of empirical cognition. For as interpreted here, empirical cognition requires bringing diverse representations together to compare, connect and judge them. And if we understand Kant to be considering this type of combination, then we see why it is not a non sequitur for him to argue as follows:
Yet the combination (conjunctio) of a manifold in general can never come to us through the senses, and therefore cannot already be contained in the pure form of sensible intuition; for it is an act of spontaneity of the power of representation, and, since one must call the latter understanding, in distinction from sensibility, all combination … is an action of the understanding, which we would designate with the general title synthesis. (B129–30, my italics)
It might seem as if a combination of a manifold obviously does come to us through the senses sometimes, e.g. when I receive a diverse array of sensory impressions all at once in the same spatial field. But Kant is not referring to such minimal combination; rather, he is concerned with our capacity to represent things ‘as combined in the object’ (B130). And he is concerned with this insofar as he assumes we have cognition. For cognition requires that we identify and differentiate objects on the basis of marks possessed by those objects.
If the combination of concern in §15 is the one involved in cognition so understood, we can see why it requires ‘an act of spontaneity’ from the understanding. For cognition requires not merely the passive intake of representations but also the capacity to be aware of one’s grounds for discriminating objects, and this requires that a subject possess a faculty of understanding through which she can give and evaluate reasons for her judgements. Such acts of judgement come from the subject herself and hence are spontaneous rather than passive.
The opening passages of §§15–16 are the strongest support for reading the B-edition’s deduction as beginning by assuming we have transcendental apperception. I have shown that these passages are more plausibly read as providing an argument – or at least the outline of one – moving from the assumption that we have empirical cognition to the thesis that transcendental apperception is required for cognition.
Summary
I have argued here for three theses: (i) empirical cognition for Kant amounts to rational sensory discrimination; (ii) an assumption of such cognition is one that Kant could have legitimately made in responding to what he understands as the Humean Problem; and (iii) the text of the Transcendental Deduction is most plausibly read as taking that assumption as its starting point.Footnote 41
