Kant's “Copernican revolution” and the “transcendental idealism” that he claimed to follow from it continue to attract some philosophers but to repel rather more. In the Preface to the second edition of the Critique of Pure Reason Kant famously wrote:
[…] up to now it has been assumed that all our cognition must conform to objects […] but let us try whether we do not get farther with the problems of metaphysics by assuming that the objects must conform to our cognition. (Bxvi)
It is an understatement to say that there is continuing disagreement about the meaning and acceptability of this proposed revolution in philosophical method. I am going to use some concepts from 20th century analytical philosophy of language to see what sense we can now make of it.
Sebastian Gardner has offered some helpful illumination of the matter in his Guidebook to the Critique of Pure Reason,[2] his main idea being that the propositions of Kant's philosophy should be understood as expressing “necessities of representation” (pp. 45-50, 304).[3] He introduces Kant's transcendental method as arising from what he rather melodramatically calls “the problem of reality”:[4]
[...] in order for reality or any part of it to become known to us, some sort of condition must obtain whereby it becomes an object for us. [...] the question is: what makes reality into an object for us? Its being an object for us is not established by its simple existence. (p. 34)
This is a generalization of the problem Kant raised in his well-known letter to Herz early in the “silent decade” of the 1770's, when he was gestating the Critique. He posed the general question: “What is the ground of the relation of that in us which we call 'representation' to the object?”[5] And about the ”intellectual” representations (which he later called the categories, or pure concepts of the understanding) he asked: “Whence comes the agreement they are supposed to have with objects?” These are arcane questions indeed, using very abstract notions of representation and of object,5 and I fear that many interpreters of Kant have not sufficiently explored their ambiguities. I suggest that if we invest close attention into how to formulate these questions about objects of representation more precisely, that may pay rich dividends.
This is a generalization of the problem Kant raised in his well-known letter to Herz early in the “silent decade” of the 1770's, when he was gestating the Critique. He posed the general question: “What is the ground of the relation of that in us which we call ‘representation’ to the object?”4 And about the “intellectual” representations (which he later called the categories, or pure concepts of the understanding) he asked: “Whence comes the agreement they are supposed to have with objects?” These are arcane questions indeed, using very abstract notions of representation and of object,[6] and I fear that many interpreters of Kant have not sufficiently explored their ambiguities. I suggest that if we invest close attention into how to formulate these questions about objects of representation more precisely, that may pay rich dividends.
Gardner poses the mind-boggling question: “What makes reality into an object for us?” (p. 34). But thinking of reality as a whole is not something that many people go in for, unless they are cosmologists, metaphysicians, or perhaps transcendental meditators! We might well wonder, on the sort of grounds Kant himself adduces in his discussion of the Antinomies, whether there can be any such thing as referring to “reality” as a completed totality. Be that as it may, it is uncontroversial that we think and talk about particular parts of reality. But there is a fundamental difference between a relational conception of representation as relating the subject to particulars existing independently of his or her mental states, and a non-relational conception of representation as the mere mental reference to “intentional objects” which may not exist outside the mind of the subject. Descartes, applying his method of doubt in his First Meditation, quickly reached a stage where he doubted the existence of the whole material world, but he never doubted that he had in his mind various "ideas", i.e. representations of things. The most fundamental question about representation can be formulated as follows: “What are the necessary conditions for us to represent in the internal, non-relational sense?” - or more succinctly: “What makes intentionality possible?”
Yet even this is still ambiguous in several ways. Firstly we must take account of the difference between two kinds of representation, namely singular reference and propositional thought.[7] In language, there is a corresponding distinction between singular terms and sentences. The former involve identifying particular objects or items (or at least the attempt to do so); whereas the latter involve propositions, which represent possible states of affairs.[8] Secondly, there is a distinction between successful and unsuccessful representations; but there are different kinds of success. The question “What are the conditions for us to represent particular parts of reality?” can be taken as about truth, or knowledge, or mere representation. On the first interpretation one would expect by way of answer a philosophical account of the notion of truth, in terms of correspondence, coherence, or a minimalist account. But talk of the truth of propositions presupposes that they have representational contents, which can be understood without being known or believed to be true. On the second interpretation the question would be about the conditions for knowledge, justification, or warrant. Whatever is necessary for representation is necessary for knowledge or belief, but the converse does not hold: we can formulate and understand many propositions for which we have little or no evidence.[9]
Questions about the success of our representations thus presuppose that they have determinate contents, so the most fundamental question of all is what constitutes representational content: i.e. what makes it the case that some of our mental states refer to, or are “about” (in the non-relational sense) particular objects or particular states of affairs? How do our thoughts have “objective reality” in the (now rather counter-intuitive) sense given to that phrase by Descartes, following the medievals?[10] As Gardner puts it, what demands philosophical explanation is “the possibility of there being objects for us, things that we can have experience of and thoughts about” (p. 37). In short, how can we think of or refer to anything in particular? It is surely this deepest level of philosophical enquiry that Kant was breaking through to in posing his questions about representation.[11]
The basic notion of success for propositional representations is truth, but it comes at the risk of the corresponding kind of failure: propositions can be false, and people can believe or assert what is false. Pace what Plato said in the Sophist, there seems to be no obstacle to expressing the content of a false belief, for that is precisely what we do in a “that-clause”, i.e. the word 'that' followed by a sentence. To be sure, there are much-discussed problems about exactly which sentences are appropriate for making reports of people’s beliefs in various contexts. But the notion of different people believing (and asserting, denying, disagreeing, or expressing agnosticism about) the same thing is one that we cannot do without, given that we are rational beings who can communicate knowledge or beliefs to each other, and raise questions about the truth of what has been said or believed by others or by ourselves at some previous time. We are thus committed to there being “objects” (in another sense of the word) of belief, assertion, denial, and mere “propounding” – and they are traditionally called ‘propositions’.[12]
A different dimension of success and failure applies to reference. Sometimes we attempt to make a singular reference, but fail to do so. Linguistically-expressed examples include the King of France in 1905’, ‘Father Christmas’, and ‘the rational square root of 2’. There are also perceptual illusions or hallucinations, as when Macbeth in his murderous state of mind thought he saw a dagger before him, or when someone takes a pattern of shadows to be a man lurking in the bushes, or a schizophrenic thinks he hears voices speaking to him. As we dangerously put it, people sometimes think or talk of, and “see” or “hear” things that aren't really there (as in the “Irish” use of the phrase ‘seeing things’). But in such cases we want to be able to explain what the mistake consists in, so we try to express what it is that the subject was thinking of, while being agnostic about its actual existence. In one sense we want to identify the “object” of their thought, while leaving it open to deny that there is any such object in another sense.
I therefore distinguish four questions within what Gardner calls “the problem of reality”:
Although distinguishable, these questions form a package, in that fully answering any one of them will involve answering the others. Reference is not a separable mental act or speech-act, it is only a preparation for saying or thinking something about what is referred to (and thereby propounding a proposition).[13] There are of course general and existential propositions which do not themselves involve singular reference; but to understand them one must be able to understand their instantiations. (1) and (3) therefore seem to be the more fundamental questions, asking how any sort of representation is possible in the first place. Yet any adequate answer must presuppose that we can make distinctions between successful and unsuccessful representations, for the very notion of representation involves the possibility of misrepresentation.[14] Talk of the “objects” of representation is thus ambiguous between two levels - what in one sense is represented whether or not the attempt is successful (the content or sense), and what is represented in the sense of being successfully referred to. We can thus distinguish four kinds of “object of representation”:
In (IIa) I have used Brentano's phrase ‘intentional object’. But this suggests that intentional objects are one kind of object that there are, a species of a larger genus of “objects”, and an overgrowth of Meinongian ontological jungle notoriously threatens this neck of the philosophical woods. Brentano described intentional objects as “inexistent”, presumably trying to avoid such ontological inflation. Some philosophers influenced by the phenomenological tradition have said mysteriously that an intentional object may or may not be identical with an actual object; but others deny that the concept of identity can apply in any such case, since intentional objects are categorically different from actual objects.[15] However I am using 'intentional object' only as a synonym for what precedes it in (IIa), namely the contents rather than the referents of mental or linguistic acts of putative singular reference. Such contents are at the level of Frege’s sense (Sinn) rather than reference (Bedeutung).
According to an influential line of thought stemming from Evans and McDowell, in a situation of reference failure there is no singular content of the act of putative reference, and hence no singular thought or proposition.[16] So in the case of someone suffering a perceptual illusion that she is being watched by a man in the bushes, she may think or say ‘That fellow is looking at me’, and we may be tempted to assume that she thereby affirms a singular thought; but if we do not ourselves believe there is anybody there, we cannot say which man she is referring to. Even if she says “That ugly little guy is looking at me”, we still cannot identify any real individual (aesthetically or vertically challenged or otherwise) to whom she is referring. There is no object in sense IIb that she is talking about, and thus no singular proposition in sense Ia that she is affirming or entertaining. I accept this analysis: I just insist that there must nevertheless be some identifiable content, albeit not singular content, of the subject's mental state at the time, for she at least believes the existential proposition that there is a man, perhaps of a certain description, in the relevant bushes. In that minimal sense, her thought has content, and that is all I mean by using the phrase ‘intentional object’ here. If that usage is deemed misleading, I do not insist on it.
The most familiar cases of (IIb) and (Ib) are of course material objects, and states of affairs involving them. These are spatio-temporally located in the physical world (albeit sometimes vaguely, e.g. clouds, mountains, tribal lands, political revolutions, and fashions of dress or speech). But (IIb) and (Ib) can also include abstract objects and facts about them, e.g. numbers, shapes and sets in mathematics, and words, phrases, and sentences conceived of as types. Many material objects and states of affairs exist independently of their being represented by anyone, or even of the existence of all human beings (though obviously this does not hold of humanly-constructed or social objects such as buildings, works of art, universities or pension funds). Philosophers of mathematics who have a realist conception of mathematical objects will want to say something similar about them. There are meaningful, contentful, putatively-referring phrases in mathematics which can be proved not to refer, precisely by reasoning from their content (e.g. ‘the greatest prime number’, ‘the rational square root of 2’). So success in representing objects of types (Ib) and (IIb) is never guaranteed - it depends on matters that are independent of the mind of the subject, and in some sense independent of all mental states of anyone.
The status of (Ia) and (IIa) requires careful elucidation. These are not representations in the sense of representings, i.e. mental acts or events such as judgings or recognitions, or longer-lasting mental states of believing. Rather they are the contents of such mental acts or states, contents that can be shared by different people, and by one person at different times. In that sense they are independent of any one mind. But they are internal objects of representings, as opposed to the mind-independent external “objects” in (Ib) and (IIb). In recent analytical philosophy a distinction has been made between “narrow” and “broad” conceptions of content:[17] only the former conception conforms to my requirements here, whereas the latter involves the existence of objects or natural kinds external to the mind of the subject. At the level of narrow content, internal objects of representation, there is no question of success or failure in representation, and no gap between thought and reality - there is merely thought. So in my narrow usage of the term ‘proposition’ in (Ia), there can be no such thing as a genuinely singular narrowly-construed proposition.[18]
Of course someone may not have acquired a certain concept, and will therefore be unable to understand a proposition or referring phrase involving it: so in that sense they can fail to represent a content of type (Ia) or (IIa). But if someone does have the concept and uses it on a certain occasion, there can be no slippage between its deployment and the existence of the relevant content. To talk of concepts is already to invoke publicly shareable standards for concept-possession and concept-application. So there is a sense in which the existence of a concept – a content or “intentional object” – with its relevant norms for correct application, though independent of any one mind, does depend on the existence of practices. These are typically the practices of minds in the plural, though we can allow the possibility of one-person practices by a Robinson Crusoe, or in any situation where someone develops a new conceptual practice which she has not yet communicated to anyone else, provided that there is no logical bar to others learning that concept, and joining in the practice of its use. Practices in this Wittgensteinian sense are essentially normative – they involve standards of correctness for concept-application. And we cannot talk of the holding of norms unless there are people (at least one person, anyway) who apply them.[19] The contents or internal objects of representation (Ia) or (IIa) are in that sense mind-dependent.
But what constitutes these mind-dependent contents? As Gardner notes (pp. 34-7), the mere existence of things or states of affairs cannot explain anyone's representation of them. From the bare fact that a certain object exists, it does not follow that any mind has any thoughts about it. And from the existence (holding true, or obtaining) of a certain state of affairs, it does not follow that anyone can understand the relevant proposition, let alone know it to be true. Obviously, we cannot give any particular examples of unidentified objects or states of affairs, for to do so would be a fortiori to identify them. But, pace Berkeley, we can quite consistently assert that there are many things and states of affairs which have not been, and may never be, individually referred to or represented by anyone, e.g. in the depths of the oceans, on other planets, in the deep geological past, or after the millionth digit in the decimal expansion of pi. And of course we believe that many of the objects and states of affairs that we do represent would still exist even if we had never done so.
For our representations to be possible, then, there must, as Gardner puts it, be “some sort of fundamental connecting relation between reality and ourselves” (p. 34). But what sort of relation is needed? In his letter to Herz, Kant seemed to think that for our empirical representations a causal relation in either direction was enough, and that his question about the possibility of representation was puzzling only for the categories, the pure or “intellectual” concepts of the understanding. But by the time he wrote the Critique, he realized that mere causation does not suffice for any sort of representation (as Gardner explains on p. 30). Even a causal relation that systematically preserves information is not enough. The variable weather of past seasons causes patterns in the growth-rings inside tree-trunks, so trees are said to contain information about past climate, and might incautiously be said to “represent” it, though trees do not have thoughts or beliefs about past climate or anything else, they do not represent in the sense we are concerned with. Even with beings like us who do represent in the full sense, there may be information about our pasts encoded in causal traces in our bodies (for example our diet can leave trace minerals in our bones), but that does not entail that we thereby represent our eating habits or know anything about isotopes. And if past physical or mental traumas cause present states of consciousness in the form of tweaks of pain or twinges of objectless anxiety, those mental events do not thereby become representations of their past causes, for no conscious memory of them need remain.
A causal information link in the other direction, from subject to object, is equally insufficient for representation. Someone may leave behind faint smells that a tracker dog can detect, or traces of DNA that can be identified with modern equipment, but that does not imply that the person has any representations of those things. Even when the causation runs from a person's mind to the world, e.g. when someone's pains or moods cause anxiety in other people, that need not involve representations of those effects (a child might feel pain, and have the concept of pain, without yet having the concept of anxiety). Of course, in our actions we intentionally bring about changes in the world, and then the subject does have representations of the state of affairs she intends to bring about. But to point to that familiar kind of intentional causation does nothing to explain how representation is possible, it just presupposes that we do indeed represent things that we intend to do.
If causal relations are not enough, what does Kant have to say in the Critique about the possibility of representation? Let us look in more detail at his programmatic introduction of his Copernican revolution in philosophy:
Up to now it has been assumed that all our cognition must conform to objects; but all attempts to find out something about them a priori through concepts that would extend our cognition have, on this presupposition, come to nothing. Hence let us once try whether we do not get farther with the problems of metaphysics by assuming that the objects must conform to our cognition, which would agree better with the requested possibility of an a priori cognition of them, which is to establish something about objects before they are given to us. (Bxvi-xvii)
This neat rhetorical contrast between “our cognition conforming to objects” and “objects conforming to our cognition” is one of those memorable turns of phrase in Kant's otherwise turgid prose that tend to dominate our philosophical imagination, so that if we are not careful we can find ourselves repeating his words without asking ourselves what they really mean. But readers of a realist disposition tend to turn away in repugnance from the very idea that objects must conform to our knowledge.[20] How, they say, can the whole universe stoop to accommodate itself to the limited capacities of our human minds; surely most of it pre-existed the evolution of human beings? Here our distinction between four uses of the term ‘object’ can be applied to relieve some of the perplexity. When Kant suggests that it has been wrongly assumed so far that our cognition must conform to objects, he is surely thinking about representation rather than truth or knowledge. For a proposition to be known, it must be true: whatever is predicated of what is referred to must “conform” to the actual properties and relations of the relevant things. For a putative singular reference to be successful, there must be an appropriate relation to an appropriate object. In that sense, Kant can and should accept that our knowledge must conform to objects of the kinds (Ib) and (IIb). He elsewhere affirms the correspondence account of truth (A58/B83, A820/B848), and that is surely part of what he means by his “empirical realism”.[21]
What then can Kant mean by his mysterious suggestion that “objects must conform to our cognition”? Not, surely, that objects of types (Ib) and (IIb) are created by our representing activities, or that even they are affected by them. Yet this notorious phrase invites some such radically idealist misunderstanding, to which a long history of Kant-interpretation bears witness. I fear Gardner is in danger of contributing to it when he writes that “in a recondite philosophical sense, the subject constitutes its objects”, and “these subject-constituted objects compose the only kind of reality to which we have access” (p. 41). This sounds dangerously idealist, unless the “recondite sense” is very carefully explained. In my understanding of Kant (and of the main trend of Gardner’s account of him) it is the contents (Ia) and (IIa) (the senses or “intentional objects” of our representations) that depend on our representing activities, for there cannot be any such contents unless there are minds with representational practices. There is a clear sense, then, in which intentional objects, the narrowly-construed contents of our representations, are mind-dependent, and can be said to “conform to our cognition”.
Let us now examine how Kant tries to follow up on his statement of his Copernican hypothesis. He applies it first to “intuition” (i.e. perceptual representation of particular objects):
If intuition has to conform to the constitution of the objects, then I do not see how we can know anything of them a priori; but if the object (as object of the senses) conforms to the constitution of our faculty of intuition, then I can very well represent this possibility to myself. (Bxvii)[22]
The next sentence (one of Kant's monstrously lengthy ones) begins:
Yet because I cannot stop with these intuitions, if they are to become cognitions, but must refer them as representations to something as their object and determine this object through them ... (Bxvii).
He has in mind here his distinction between intuitions and concepts, sensibility and understanding, and the corresponding textual division between the Transcendental Aesthetic and the Transcendental Logic. In the former he proposes to examine “our faculty of intuition” before dealing with concepts and conceptual cognition, which involve our “faculty of understanding”. However, in view of his subsequent statement that intuitions without concepts are blind and that only through the union of sensibility and understanding can cognition arise (A51/B75-6), we may wonder how it is possible to discuss human perception without bringing in conceptualization. Interpreters have come to agree that Kant's notion of “intuition” (Anschauung) is ambiguous between non-conceptualized and conceptualized representations. There is unconceptualized perception in animals and pre-linguistic infants, and it is also manifest in human adults, for instance in playing ball games, in our sensitivity to facial expressions and tones of voice, and in our appreciation of music and abstract art. Whether it count as representation depends on how we decide to use the word ‘representation’ as a technical term, but there seems to be no harm in extending the word to non-conceptual cases (as Kant himself does[23]), providing we recognize the differences between them and conceptual representations. Of course, we cannot say anything about objects of non-conceptual representation without using concepts: but that is not to attribute those concepts to the relevant subjects. We know that a lamb can see, hear, or smell things, and we can tell when it recognizes (or misrecognizes) a sheep as its mother, without thereby our having to credit with the concept of sheep or of motherhood. There is a wide Aristotelian sense of the terms ‘experience’ and ‘mentality’ in which experiences, mental states, and representations can be ascribed to any creature capable of unconceptualized perception, desires, emotional arousal, and perception-guided behaviour.
Kant, however, has a narrower notion of experience (Erfahrung) which is peculiar to human beings. This involves the subject applying concepts to represent objects and states of affairs, to make perceptual judgments and to act on them, and to evaluate the reasons or justification for judgments and actions - and all this can be expressed by the subject in language. Our most distinctively human representations are thus conceptual as well as perceptual. But not all our representations are conceptual, we have many perceptions which remain unconceptualized. And not all humans have any conceptual representations: infants lack them, and those affected by brain damage or degeneration may lose them. We can thus agree with Kant in saying that it is a synthetic yet in some sense a priori truth that human representation – i.e. our typical, mature form of representation – is conceptual.
The above discussion is not confined to presently perceptual representations, namely those which involve the subject's own current perceptions. Any general account of representation must also apply to our knowledge of the world in geography, history, and the natural and social sciences, some small part of which may be arrived at by induction from one's own past perceptions, but most of which is acquired from testimony, by believing (and in favorable cases) coming to know what others inform us about. The undeniable core of empiricism is that all knowledge of contingent facts about the material world must be perception-based, in the sense that it ultimately depends for its justification on perception by somebody or other. Yet this does not rule out the conception of a “God's-eye-point-of-view”, not itself in space or time, yet enjoying representations of the physical world which are not perception-based; theologians have sometimes pictured God's omniscience in such a way, and it is not obviously incoherent. Kant himself maintained the logical possibility of a non-perception-based representation of reality, which he called “intellectual intuition” (B71-2, B145, B308). This would not involve any causal affection by its objects, and it would be free of the subjective features and limitations that characterize our perception. It might therefore be said to represent things “as they are in themselves”, independently of the perceptual conditions that apply to our human representation. But Kant admits that since our intuition is “sensible” we cannot know what it is like to possess this alleged intellectual intuition. Some philosophers may want to rule it out as impossible - but it is not clear what the limits of possibility or conceivability are here. Short of a knockdown proof, perhaps we can agree with Kant in saying that it is a synthetic a priori truth that all human representations of the spatio-temporal world are perception-based.
What sense can we make, then, of Kant's somewhat shocking suggestion that objects “conform to the constitution of our faculty of intuition”? Let us distinguish the following conceptualized perceptual subspecies of our previously listed kinds of ‘object’:
There is no question of objects in senses (Ib-cp) and (IIb-cp) “conforming to the constitution of our faculty of intuition”, or to the perceptual sensitivity of any particular creature or species. But talk of (Ia-cp) and (IIa-cp) only makes sense as a way of talking about the perceivings and perceptual sensitivities of rational beings like ourselves. The contents or intentional objects of our perceptual states must of course “conform to”, or depend on, the ways in which we can perceive. We know that the intentional objects of human perception have certain properties: we know, for example, that everything we can see is colored within certain wavelengths of light, and that the sounds humans can hear lie within certain frequencies whereas dogs and bats can hear more, and dogs can smell much more than us. This, however, is a posteriori psychological knowledge, resting on the contingent differences between the sense-organs of various species. But when Kant asserts in the Transcendental Aesthetic that space and time are features of “the constitution of our faculty of intuition”, he is claiming that the spatio-temporal form of our intuition is a deeper-lying truth about us, involving a stronger kind of necessity than biological facts about our human sense-organs. It is surely a necessary truth that all finite rational beings have conceptualized perceptual representations, from their positions in space and time, of things and events in the physical world.
Let us now examine the next part of that lengthy sentence that I interrupted above:
Yet because I cannot stop with these intuitions, if they are to become cognitions, but must refer them as representations to something as their object and determine this object through them, I can assume either that the concepts through which I bring about this determination also conform to objects, and then I am once again in the same difficulty about how could know anything about them a priori [...]. (Bxvii)
As we have acknowledged, there is unconceptualized perception in humans, but our empirical knowledge involves the application of concepts to our perceptions. The logical form of singular perceptual judgments is ‘This F is G’, in which the concept F is used with a demonstrative word (and sometime a gesture) to make identifying reference to a presently perceived object (of type IIb-cp), and the concept G expresses a property which the subject claims to presently perceive as holding of that object. We can also say that the subject currently perceives the corresponding fact or state of affairs (of type Ib-cp), namely that the indicated F falls under the concept G.
Kant's claim in the second part of the quoted sentence is that the notion of “concepts conforming to objects” cannot explain how a priori knowledge is possible. But what can it mean to say that concepts “conform” to objects of type IIb-cp? Concepts can apply to things, of course. And natural kind concepts can be said to aspire to "conform to objects" in a deeper sense, aiming to "carve nature at the joints", i.e. to make the classifications which will prove to be most explanatory in the scientific sense. Kant touches on this topic much later on in the Critique, in the Appendix to the Dialectic, where he touches on the philosophy of science (A642/B670ff), and also in the Method section when he discusses the possibility of definitions (A727/B755), but he must surely admit that finding out which natural kind concepts apply to the world – i.e. discovering natural kinds – is an empirical (a posteriori) matter. So there is after all one sense, consistent with the rejection of “transcendental realism”, in which the existence of objects falling into certain natural kinds or species can explain why humans have acquired concepts of those natural kinds. Interaction with certain kinds of object, e.g. the repeated perception of iron, oak trees or elephants, and indeed the practical use of them (Heidegger's Zuhandensein – ready-to-handedness) is part of the explanation of how human communities came to form concepts of those kinds of thing. But that is empirical historical explanation of the development of specific concepts by certain peoples; whereas we were seeking an explanation at the philosophical level of how any conceptual representations at all are possible. So let us proceed to the positive horn of Kant's dilemma:
[...] or else I assume that the objects, or what is the same thing, the experience in which alone they can be cognized (as given objects) conforms to those concepts, in which case I can immediately see an easier way out of the difficulty, since experience itself is a kind of cognition requiring the understanding, whose rule I have to presuppose in myself before any object is given to me, hence a priori, which rule is expressed in concepts a priori, to which all objects of experience must therefore necessarily conform, and with which they must agree. (Bxvii)
This passage is, no doubt, one of those which make many readers give up hope of illumination from Kant! For one thing, it is puzzling how he can identify objects with experience of those same objects. But if we deploy once more our distinction between intentional and actual objects, we can offer a plausible reinterpretation. For Kant, “experience” (Erfahrung) means conceptualized perceptual experience, expressible in perceptual judgments involving the faculty of understanding (Verstand) as well as sensibility (Sinnlichkeit). In terms of the distinctions made in this paper, we can take his point to be that the narrowly-construed contents or intentional objects of experience, (Ia-cp) and (IIa-cp), must “conform” to a priori concepts (the categories, or “pure concepts of the understanding”), which express rules to which all of our conceptualized perceptual experience must conform. We have here in the B Preface a sneak preview of the kind of argument Kant offers in the Transcendental Deduction and the Analogies of Experience. Strictly speaking, we cannot literally identify the actual objects with the intentional objects, nor the latter with conceptualized perceptual experiences, but the intentional objects can be said to be the content of such experiences.
Let us now begin to explore how this preview fits with the main body of the Critique, especially those passages where Kant explicitly discusses “objects of representation” in the Transcendental Deduction. In the A version he poses the question: “What does one mean, then, if one speaks of an object corresponding to and therefore also distinct from the cognition?” (A104). In trying to answer this question, we must be careful about the use of the first-person plural (quite often in philosophy, we need to ask “Who exactly are we?”). One person A can distinguish between the objects he believes to exist, and an overlapping but non-identical set of objects that (he takes it) another person B recognizes – some of which A may not believe to exist, and will therefore describe as mere intentional objects of B's thought. A can similarly distinguish between the set of objects he now recognizes, and those that he believed in at some stage in the past. Given two groups of people the Cs and the Ds, each with a shared set of beliefs, the Cs can distinguish between their own set of intentional objects of thought and those of the Ds (e.g. we believe in mental illness, where they believe in demon-possession), and the Cs may also distinguish their present set of objects from those they recognized in the past (e.g. we used to believe in phlogiston, but now we believe in oxygen). But of course no individual or group can make a distinction between their own current set of intentional objects and the objects they currently believe to exist. I suggest this is why Kant answered his own question as follows:
It is easy to see that this object must be thought of only as something in general = X, since outside of our cognition we have nothing that we could set over against this cognition as corresponding to it (A104).
But this is a dangerously ambiguous statement. Read in one way, an idealist abyss threatens to open up, for we are tempted to picture our thought as inevitably but frustratingly cut off from direct contact with the world, so that any knowledge of mind-independent reality comes to seem impossible. But surely this is a philosophical illusion. It is undeniable that if one is to refer to and describe objects one must use some way of doing so do: there can be no such thing as thinking about a particular item without relying on some humanly-usable procedure for identifying exactly which item one “means”, i.e. intends to refer to. Thus there can be no reference without sense, if “sense” is understood with Frege as a way of singling out the item referred to. (There is almost always more than one conceivable way of singling out the same thing.) So it seems that the notion of an object distinct from all our ways of referring and knowing can only be of something unspecified “thing in itself”, like the unknown quantity x in an algebraic equation. The very talk of an object presupposes that one particular object, entity, substance, event, process or state of affairs is already somehow identified determinately and discriminated from all others. But identification (singling out, discrimination) is something that we have to do in order to represent any particular aspect of reality. The notion of reality “as it is itself”, prior to and independent of all our representing activities, can only be expressed by all-encompassing, indiscriminate, un-particularizing words like ‘Reality’, ‘the World as it is in itself’, ’Nature’, ‘Being’, or ‘the One’ – with a hint of mysticism!
Is there really any such dramatic threat to our sense of objectivity, our assumption that our thoughts (or some of them, at least!) refer to mind-independent reality? Although one person cannot distinguish between her present set of intentional objects and the objects she presently believes to exist, she can conceive of changes in her beliefs – and the same is true of human communities. We had better allow the possibility that our set of intentional objects and accepted propositions will expand, and occasionally contract, as we gain more knowledge of the world. There are many ways in which present beliefs can be tested by further experience, involving further perceptions by oneself, the testimony of other people, or induction and scientific inference generally. Kant's next sentence reads:
We find, however, that our thought of the relation of all cognition to its object carries something of necessity with it, since namely the latter is regarded as that which is opposed to our cognitions being determined at pleasure or arbitrarily rather than being determined a priori, since insofar as they are to relate to an object our cognitions must also necessarily agree with each other in relation to it, i.e. they must have that unity that constitutes the concept of an object. (A104-5)
Our thought of the relation of our present theories to their objects carries with it an element of necessity, in the sense that there are necessary connections between our present beliefs and many possible perceptions, whether our own or other people's. Our beliefs about what there is in the world, including our perceptual judgments, cannot be totally haphazard or arbitrary, having no connection with one another: if various beliefs are to be about particular objects within the world, they must be consistent with each other. And perceptual judgments about an object must be interpreted as having a causal relationship with that object, i.e. perceptual confrontations with it must have caused the perceptual experiences on which those judgments were based.
It thus emerges that representation is essentially holistic. Referring to a particular object – or having a singular thought – is not a simple property which a given mental state either possesses or lacks quite independently of what holds true of everything else in the mind of the subject. A given mental state represents an object only in virtue of its conceptual role, which depends on a complex pattern of actual and possible relationships to other representational mental states in that subject (and arguably in other people too, according to Wittgenstein’s argument against the possibility of a “private language”). As Kant goes on:
It is clear, however, that since we have only to do with the manifold of our representations, and that X which corresponds to them (the object), because it should be something distinct from all our representations, is nothing for us, the unity which the object makes necessary can be nothing other than the formal unity of the consciousness in our synthesis of the manifold of our representations. Hence we say that we cognize the object if we have effected synthetic unity in the manifold of intuition. (A105).
The same theme is expressed in slightly different terms in the B Transcendental Deduction when Kant says that an object “is that in the concept of which the manifold of a given intuition is united” (B137).
Towards the end of the A version Kant sums up, in terms that he himself admits are paradoxical, the idealist-sounding conclusion to which he thinks his argument has led:
Thus we ourselves bring into the appearances that order and regularity in them that we call nature, and moreover we would not be able to find it there if we, or the nature of our mind, had not originally put it there.
[…] Thus as exaggerated and contradictory as it may sound to say that the understanding is itself the source of the laws of nature, and thus of the formal unity of nature, such an assertion is nevertheless correct and appropriate to the object, namely experience. (A125, 128)
Two centuries later Leszek Kołakowski (writing in the very different context of Polish communism) expressed a similar thought in words which also seem “exaggerated and contradictory”:
The picture of reality sketched by everyday perception and by scientific thinking is a kind of human creation (not imitation) since both the linguistic and the scientific division of the world into particular objects arise from man’s practical needs. In this sense the world’s products must be considered artificial. In this world the sun and the stars exist because man is able to make them his objects, differentiated in material and conceived as ‘corporeal individuals’.[24]
But the whole trend of this essay suggests that we should not rest content with these dangerous formulations. David Wiggins in his lifelong study of identity and individuation has wisely insisted that we need to walk a philosophical tightrope between the errors of idealism on one side and non-conceptualist realism on the other. We must reject any ‘sense’ in which horses, leaves, sun and stars could be supposed to be artifacts, while acknowledging that we have to use our concepts in singling out those particular things with their criteria of identity, amongst many others that could be singled out by other concepts:
Conceptualism properly conceived must not entail that before we grasped [the relevant] concepts, their extensions did not exist autonomously. […] Its most distinctive contention is that, even though horses, leaves, sun and stars are not inventions or artifacts, still, if such things […] were to be singled out in experience at all so as to become objects of thought, then some scheme had to be fashioned or formed, in the back and forth process between recurrent traits in nature and would-be cognitive conceptions of these traits, that made it possible for them to be picked out.[25]
Nicholas Rescher has also defended “conceptual idealism” in his book of that name,[26] with more emphasis on the way in which laws and unamplified possibilities enter into our conceptions of things.
At the end of the B Transcendental Deduction Kant himself offered an account that is perhaps rather less “exaggerated and contradictory” of how the categories can prescribe laws a priori to appearances:
It is by no means stranger that the laws of appearances in nature must agree with the understanding and its a priori form, i.e. its faculty of combining the manifold in general, than that the appearances themselves must agree with the form of sensible intuition in general. For laws exist just a little in the appearances, but rather exist only relative to the subject in which the appearances inhere, insofar as it has understanding, as appearances do not exist in themselves, but only relative to the same being, insofar as it has senses. (B164)
Here is an interpretation of this passage, in the light of the distinctions deployed in this essay: just as all the objects of unconceptualized perception (“appearances” in one sense of the word) must conform to the conditions for the subject to have perceptions, so all the objects of conceptualized perception (“appearances” in a conceptualized sense – Ib-cp and IIb-cp) must conform to the conditions for the subject to have concepts of perceivable objects, properties and states of affairs.