In her new book, Manifest Reality, Lucy Allais incorporates many of her thoughts on Kant’s views about sensory intuitions and concepts into a systematic interpretation of Kant’s theoretical philosophy as a whole. Allais is well known for taking the ‘non-conceptualist’ side of the conceptualism/non-conceptualism debate. As the debate is sometimes framed, conceptualists maintain that it is impossible to have intuitions of objects independently of the employment of concepts, while non-conceptualists maintain that intuitions of objects without concepts are possible. Thus construed, the debate is about what it takes for singular representations to have objective purport, or to be ‘about’ something in the world. But the debate is also sometimes understood as one about whether our human capacity to represent objects in intuition depends on the capacity for conceptual representation, as conceptualists maintain, or whether the same sensory capacity could be found in animals that lack conceptual capacities, as non-conceptualists maintain. Allais is a non-conceptualist in both senses. She cashes out the independent contribution that sensibility makes to cognition both in terms of the possibility of sensible intuitions that are not (yet) conceptually determined (pp. 173ff.) and in terms of the possibility of having our a priori forms of intuition (and hence a faculty for representing things in intuition) without having a faculty of concepts (p. 175). As Allais notes, the latter form of non-conceptualism is compatible with thinking that ‘a non-concept having animal that has a spatial form of intuition does not represent space as the unified object that is the object of study of geometry’ (p. 175). One might think that non-concept bearing animals share our forms of intuition and still think that they are incapable of cognition, including geometrical cognition, since they lack concepts and concepts are required for cognition of objects.
I think there are good reasons to accept the first and to reject the second form of non-conceptualism (since non-rational beings cannot share our forms of intuition). However, in this paper, I want to shift our attention to a slightly different issue. I wish to reformulate the conceptualist debate as a debate about knowledge, which was McDowell’s main focus when he first formulated his classic version of a conceptualist reading of Kant. We can begin with the Kantian claim that knowledge involves two components:
(1) The object of knowledge must be given to me in intuition.
(2) That which is given must be subsumed under concepts (including the categories) in thought or judgement.
The question I wish to ask is whether these two components are separable in knowledge; my conceptualist will maintain that they are not, while my non-conceptualist maintains that they are. The debate here does not concern the question whether these components are separable outside of knowledge. That is, I do not want to question Allais’s claim that objects can be given in intuition independently of concepts, as when we are merely associating representations or perceiving something we cannot grasp or understand. The controversy is instead one about knowledge only: assuming that I have knowledge of an object and am not merely associating or ‘playing with’ representations, are these two elements in knowledge (intuition and concept) separable? That is, can we make sense of how the object of knowledge is given to me in sensibility independently of how it is being thought through the categories? To think of these components of knowledge as inseparable is not to deny that their roles are ‘distinct’, or that they each make ‘an essential contribution to cognition, and that neither can replace the other or play the other’s role’ (p. 145). But it is to deny that these distinct roles can be understood apart from one another, i.e. that sensibility can play its distinctive role of giving an object in an act of knowing independently of the understanding’s role of thinking, and vice versa. For instance, you might think that sensibility can play its role in giving an object through the schema of persistence only in virtue of the object’s being thought through the category of substance.
Allais may object to this focus on ‘knowledge’, since Kant seems to be more concerned with the contributions that sensibility and understanding make to ‘cognition’ (both are possible translations of Erkenntnis). Whereas knowledge is factive, cognition (as that term is used by interpreters of Kant) can be true or false. However, Allais does concede that Kant’s project can be understood as one of establishing the possibility of metaphysics (p. 5). This must mean that he wishes to show how a priori, metaphysical truths about the world are possible – such as the truth that ‘all intuitions are extensive magnitudes’ (Axioms of Intuition B202), or that ‘all alterations occur in accordance with the law of the connection of cause and effect’ (Second Analogy B232), or that ‘that which is connected with the material conditions of experience (of sensation) is actual’ (second Postulate B266).Footnote 1 And it is surely one of Kant’s (Copernican) insights that one cannot show how these truths are possible independently of showing how knowledge of them is possible. If he can only tell us how it is possible to think or ‘cognize’ these propositions, regardless of whether or not they are true, he would not have shown how the truths themselves are possible (our minds may be able to think what cannot possibly be true). So if he establishes the possibility of these truths, his investigation cannot stop short of showing how metaphysical knowledge (and not mere cognition) is possible.
In the following, I will first show how Allais is committed to a separability claim, or to non-conceptualism about knowledge in the above sense (§1). I then argue that Allais’s interpretation rests on what McDowell calls a ‘hybrid conception of knowledge’, and I will try to show that her attempt to extricate herself from the problems associated with a hybrid conception of knowledge does not work (§2). This will be shown to have consequences for the kind of idealism that Allais can ascribe to Kant (§3).
1
As noted above, Allais does not deny that concepts are required for knowledge (or ‘cognition’). More specifically, knowledge involves syntheses of representations that are governed by concepts (the categories). What is controversial is how concept and intuition come together in knowledge. Allais says this about their relation:
The conceptualist reading sees Kant as saying that categorically governed synthesis is necessary to produce intuitions – presentations of unified, single particulars. However, the syntheses Kant argues for in the Deduction are something that is done to intuitions, and therefore they need not be understood as what produces intuition, or as something that is necessary to have intuitions. (p. 169)
Allais’s assertion that the ‘syntheses’ that are at issue in the Deduction – syntheses that are governed by the categories – are ‘done to intuitions’ admits of various interpretations, and focusing on these will I hope bring out more clearly what I mean by characterizing her position as one that involves a ‘separability’ claim.
Artefact model
Syntheses are done to intuitions as a craftsman ‘does something to’ materials in producing an artefact, such as a table. The craftsman imposes the form of a table onto given matter. With regard to the matter, the same wood that serves as matter for this table could have been made into something different; the wood in the table would still be wood when taken out of the table. There is, therefore, a certain degree of arbitrariness in the fact that this wood is informed by the form of a table. Transferred to the case of conceptually governed synthesis, this would mean that it is arbitrary that our sensible manifold is ‘informed’ by the categories; the manifold might not have been informed in these ways. For instance, the way in which the object is given to me in intuition would be the same, even if I did not think of the object as a substance (just as the wood still is wood, even when it is taken out of the table). Moreover, the model suggests that the form too could have been different. Instead of the form of a table, the craftsman could have imposed the form of a chair onto this wood. Analogously, although our sensibility is informed by the categories in acts of knowledge, it could have been informed by a radically different conceptual scheme.
Food Model
Syntheses are done to intuitions as an organism ‘does something to’ materials in its environment when it eats. The materials must be ‘informed’ by the life-form of the organism to become part of the organism. Once they are eaten by man, for instance, they become part of the body of man, and so are no longer ‘separable’ from man in the way that the wood is separable from the table. (Whereas the wood remains wood when separated from the table, blood is blood in name only when fully separated from its function in the body.) On this reading, the categories determine the way in which objects are given to me in sensibility; once I have knowledge of an object as a substance, it is given to me differently in intuition.
Organ Model
Syntheses are done to intuitions as the life-form of an organism informs its organs (parts of the body). Apart from the informing activity (life activity) of the whole organism, the bodily organs each would not be what they are, since they have their functions only in relation to the function of the whole. On this reading, intuitions cannot perform their function of giving objects at all independently of the understanding’s function of thinking them through the categories.
The first model is committed to what I am calling the ‘separability’ of intuition and concept in knowledge. Knowledge, on the artefact model, is a loose ‘unity’ of intuition and concept, because the same materials in the whole remain what they are even in separation from this whole. On this model, the understanding (through the categories) does not affect the way in which objects are given in intuition. But it does add a function to the intuitions that was not there before: their function in knowledge (just as the wood acquires a function in tables once it is formed into a table). This seems to be how Allais understands the relation between sensibility and understanding in knowledge, since she thinks that the Deduction shows only that the categories are conditions for thinking, and thereby knowing, objects, but not for the objects themselves, or for their being given to me in sensibility:
Kant opens the Deduction saying that the categories are not necessary for anything to be given to us in intuition, and he equates showing that without the categories nothing is possible as ‘object of experience’ (A93/B126) with showing that the categories are necessary in order to think [Allais’s emphasis] objects. Thus, when he talks about what is necessary for something to be an ‘object of experience’ or to ‘be an object for me’, he is not talking about what is necessary to intuit an object but about what is necessary for us to be in a position to cognize an object and, specifically, about the conditions of one aspect of cognition – concept application or thought about objects. (p. 263, emphasis added)
2
I think there are both philosophical and textual grounds for calling Allais’s separability thesis into question. The philosophical grounds relate to the dangers of what McDowell has called a ‘hybrid conception of knowledge’. In the hybrid conception, ‘a satisfactory standing in the space of reasons is only part of what knowledge is; truth is an extra requirement’ (McDowell Reference McDowell1995: 884). The first condition for knowledge is satisfied by the subject’s own exercises of spontaneity in thought and judgement (or in McDowell’s terms, by her epistemic agency in the responsible formation of judgements) – this is what makes one a satisfactory epistemic subject. But on this view actual truth is not guaranteed by one’s satisfactory standing in the space of reasons; truth is an additional, external component of knowledge (a gift from the world). McDowell argues that, on a hybrid conception of knowledge, it is in a significant sense accidental that one’s responsibly formed belief is true. For the same position in the space of reasons could be occupied by someone whose belief is false. This contradicts the concept of knowledge, which must be distinguished from merely true belief or lucky guessing. If I really know that S is P, my belief that S is P must be non-accidentally true, in the sense that it is an achievement – it is true not from luck, but because of my standing in the space of reasons. Or in more Kantian terms, my judgement is true of the objects because it is a successful exercise of a capacity for knowledge. Of course, much more would have to be said about the relevant sense of ‘because’ here, but the relevant sense – whatever it is – would distinguish judgement that is knowledge from accidentally true judgement.
One indication that Kant holds a non-hybrid conception of knowledge is that he says that when the understanding (capacity for knowledge) operates in accordance with its own inner laws, the result is knowledge. A satisfactory exercise of the capacity for knowledge (under sensible conditions) is sufficient to generate knowledge, and does not need to await a gift from the world. ‘Hence ... the understanding by itself (without the influence of another cause) . . . cannot [err], . . . because while it acts merely according to its own laws, its effect (the judgement) must necessarily agree with these laws’ (A294/B350; see also Logik 9: 53). The cause of all error therefore must lie in something external to the understanding: the ‘unnoticed influence of sensibility on understanding’ (A294/B350).
How is the hybrid conception of knowledge connected to Allais’s separability claim? Empirical judgements are non-accidentally true only if the objects of sensibility non-accidentally agree with them, or if this agreement of the cognition with the object (which Kant says is the nominal definition of truth) is due to an exercise of the capacity for knowledge. So it must be shown that the ways the objects appear or are given in experience non-accidentally agree with the ways they are thought or judged. But if Allais’s separability claim were correct, the objects of sensibility would only accidentally agree with the forms of judging them (the categories). Whether the objects of sensibility agree with my judgements and whether my judgements are competent exercises of the understanding are two wholly separate matters (the former is not due to the latter). Allais’s interpretation thus commits her to a hybrid conception of knowledge, which threatens to make knowledge indistinguishable from merely true belief.
Allais appears to have this sort of objection in mind when she invokes the argument from the Aesthetic for transcendental idealism in her reading of the Deduction. The Deduction on its own, according to Allais, establishes a merely conditional claim: the categories apply to objects of sensibility, insofar as these objects are subsumed under the categories in thought and judgement (pp. 291–2). For Allais this means that the argument is epistemological: it establishes that the categories are necessary for knowledge and thought of objects, but not that they are necessary for the objects themselves. ‘The problem is not with what it takes for something to be an object (to exist) but with what it takes for us to be in a position to think of anything as an object’ (p. 285). This still allows for the possibility that the categories do not apply to, or are not true of all spatio-temporal objects, since it is not the case that all spatio-temporal objects are known or thought. According to Allais, we are entitled to extend the application of the categories to all spatio-temporal objects only under the assumption of a previously established idealism about appearances:
The categories can be known to apply to all spatio-temporal objects not because they are conditions of being given objects in intuition (they are not ways of carving particulars out of a sensory mass) and not because they are necessary for thought to have relation to an object (to have determinate objects of thought to which empirical concepts can be applied), but because spatio-temporal objects are limited to the conditions of our cognition and the categories are conditions of our cognition. (pp. 290–1)
Kant is able to go beyond a conditional claim about a priori concepts being necessary conditions of empirical concept application to the claim that spatio-temporal objects are all subject to the categories because he has already shown that spatio-temporal objects are mind-dependent appearances that are limited to the conditions of our being able to cognize them. (p. 299)
Allais here introduces – illicitly, in my view – the general notion of ‘conditions of our cognition’ into the conclusion she draws from the Aesthetic. The Aesthetic, on Allais’s own reading, shows that ‘something is empirically real only if it is possible for it to be presented to us in an intuition’ (p. 207), or as she says elsewhere, that ‘the object presented does not exist independently of the possibility of its being present to consciousness’ (p. 201). All that the Aesthetic can establish on Allais’s reading is that the objects of intuition (intuiteds) must be possibly given in intuition. It is a slide to go from this claim to the more general claim that they must be possibly cognized, or are limited to conditions of cognition, and therefore must be possibly thought or known through the categories. Only if she makes this slide to the possibility of their being thought (or judged) can she conclude that the categories apply to all objects of sensibility.
Allais has transposed the problem of the unity of concept and intuition to the metaphysical realm. We now have objects for which the following holds:
(a) they are real (or exist) only if it is possible for them to be given to us in intuition;
(b) they are real (or exist) only if it is possible for them to be thought through the categories.
But why is it that objects that can be intuited also are things that can be thought (especially since their being intuited does not depend on their being thought)? Why is it not a mere accident that (a) and (b) hold for the objects? And if it is a mere accident at the metaphysical level, this still leaves us with the problem of the accidental relation between intuition and concept at the epistemic level. The problem facing Allais’s hybrid conception of knowledge does not go away by being transposed to a metaphysical level. In fact, the problem intensifies. For how should we ever establish a necessary connection in the objects themselves if this is a connection that holds independently of the act of knowing (or thinking) them? This would be like stepping outside of ourselves to establish what is true about the objects independently of the act of knowing what is true.
3
If Kant is committed to Allais’s separability claim and to a hybrid conception of knowledge, this has important consequences for Kantian idealism. In particular, it seems to commit Kant to precisely the kind of phenomenalist idealism that Allais wants to avoid ascribing to him. According to Allais, phenomenalism is the view that ‘empirically real, spatial objects are things that exist only in the mind ... or that they exist as constructions out of mental items or mental states or mental activities, or that they supervene on properties of mental states’ (pp. 37–8). But if the separability claim is right, the materials of sensibility could have been informed by a very different conceptual scheme in a different kind of being – or at least, there is nothing in the relation between sensibility and the understanding, as conceived by the artefact model, that rules out this possibility. So we are not entitled to the claim that the world as we think it (through our categories) is the world as it is objectively, or independently of how we happen to think it, but can only say that it is as it is for us who think it in this manner. For the world as it is thought is a world that our understanding has ‘imposed’ its forms onto, or that is informed by our ways of thinking. So the world as we think it – and thus, the world cognized by us – would be a world that exists as a construction of our mental, intellectual activities. Kant’s idealism would thus emerge as a form of (what Hegel called) ‘subjective idealism’, or what Allais has called ‘phenomenalism’.
Allais could respond by saying that this places too high a bar on objectivity. We cannot demand that objective knowledge requires knowing how things are independently of our acts of knowing or thinking them – that would be like demanding that we jump outside of our skins to test whether what we think is how things really are. But the point is that the artefact model requires that we think of a world external to thought as possibly different from how we think it. For if the understanding imposes forms onto sensible representations as a craftsman imposes forms onto his materials in producing an artefact, then it is only accidental that the things given in sensibility are thought through our categories. And this is just another way of saying that we cannot rule out the possibility that these things might not have been thought through these concepts of the understanding, and thus that our thoughts that apply the categories to objects, if true, are only accidentally true. What we can claim to know, where knowledge implies non-accidentally true belief, is only that the world as it is for us (as it is represented in our thoughts) is categorically structured.
Notice that Allais also cannot say that, although the world as it is thought or cognized is an ideal world, the world as it appears in sensibility is real. For Allais recognizes (I think correctly) that Kant’s transcendental idealism hinges entirely on the ‘ideality’ of our forms of sensibility: there is no guarantee that the world as it appears is the world as it is objectively, because there is no way to rule out the possibility that it could have appeared differently, through different forms of sensibility. It seems that Allais is only allowed to think of things in themselves as objectively real, a position that is compatible with phenomenalism as defined above.
As I mentioned above, I think Allais is right, against McDowell, to emphasize that intuitions are possible that are not actually ‘informed’ by concepts of the understanding. We do not need to think of the categories as necessary for all sensible singular representations of objects, as on an ‘organ model’ of their relation. But this amendment does not affect McDowell’s objections to the hybrid conception of knowledge. McDowell is right, against Allais, that in knowledge the way objects are given in sensibility must be made possible by the categories (as on the food model). Only if Kant succeeds in showing that the understanding informs sensibility, not as a craftsman imposes forms on independently given materials, but as the life-form of an organism ‘informs’ the materials given to it from elsewhere, will he succeed in showing that it is non-accidental (in the relevant sense) that the categories apply to (or are true of) the objects of sensibility, and thus that knowledge of a categorically structured world is possible.
