Introduction
There are a number of forks in the road that interpreters of Kant encounter. For example, there is the choice about Kant’s notion of the unity of apperception: is it essentially a unity that refers to the self and its states, or something else, perhaps the unity of the proposition? Landy chooses the first fork, but with little in the way of argument against the second. There is the choice to take transcendental idealism seriously as Kantian ontological doctrine or to take it in some other way. Landy takes it in some other way. There is a choice about Kant on the arrangement among, and representational power of, intuitions: do both exist in virtue of some intrinsic feature of intuitions or do they (one or both) arise as a result of the imposition of conceptual forms on intuitions? The strongest conceptualist position is that both the arrangement and representational power depend on conceptual imposition. Landy takes a strong conceptual stance. Finally, there is the choice whether to take Kant’s doctrine of the imagination as constructor of images literally and seriously. Like many contemporary interpreters, he does not.
Landy’s main objectives here are (1) to develop a dialectic between Kant and Hume on the nature of the representation of complex objects, and on the possibility of having ideas of mind-independent objects, of natural necessity, and of the self, which Kant wins in all four cases; (2) to defend a strong conceptualist account of intuitional representation; (3) to propose an interpretation of Kant’s doctrine of the ‘unity of the proposition’ in terms of inferential commitments; (4) to construct a detailed model, based on Sellarsian ideas, of how the arrangements and structures of the sensory manifold are caused by our conceptualization of a world of objects and how our sensations thus structured act as an empirical constraint on said conceptualization; (5) to offer an account of a Kantian theory of theory-change in science, including theory-replacement, as occurring within Kant’s theory of experience rather than his theory of noumena, as Sellars maintains; and (6) to offer a sustained defence of the formalist reading of Kant’s doctrine of the self, primarily as it occurs in the Paralogisms. Landy also offers (7) an interpretation of the Metaphysical and Transcendental Deductions. His reading of the latter (p. 112) makes the assumption that the analytic unity of apperception is per se about the identity of the self and is represented in the principle D2: [The I that thinks x] = [the I that thinks y] = [the I that thinks z].Footnote 1
Assessment
Depending on whether one takes it to be a just-so story or to reflect an actual engagement of Kant with Hume, I take the first objective to be quite successful, and not so successful, respectively. The second objective is achieved as well as anyone could, given a strong conceptualist background, and the same is true of objectives (3) and (4), especially (4). Objective (5) seems to me to also be very successfully achieved. Regarding objective (6) my assessment is, as before, that it is the best that can be done from a conceptualist point of view, which is very good indeed, given the complexities and obscurities inherent in these texts. Regarding (7), Landy’s readings of the Metaphysical and Transcendental Deductions are clearly laid out – some of the clearest that I have seen – detailed and plausible. For a reader who likes Sellars and Kant and good, jargon-free Kant-exegesis combining philosophical plausibility and exegetical novelty, this book is highly recommended.
Landy’s central contention is that concepts are rules of inference. Since rules of inference take propositions as inputs and outputs, and rules for constructing images do not, this means, pace Landy,Footnote 2 that for Landy, Kantian concepts are not rules for constructing images. My main complaint falls on this contention. Refusing to fold image-construction rules into the mix leaves Landy without an adequate reading of Kant on the unity of the proposition and on the issue of the empirical constraint on judgement. Or so I argue in the final section. The argument will rest on a non-conceptualist reading of Kant on intuitions and so constitutes a response to a challenge Landy throws in the non-conceptualist’s direction.Footnote 3 Next up is an outline of the main trajectory of Landy’s argument.
The Main Trajectory
Landy finds in Hume a distinction between ideas that picture a content and ideas that represent a content. An idea pictures a content when it contains the content as a copy of an impression that contains that content. Hume’s famous Copy Principle says that no idea picturing x arises unless it comes from an impression which contains x. But Landy also finds that Hume needs a second principle to allow an inference from picturing ideas to representing ideas. This is the Representational Copy Principle. ‘Hume uses it to move from, for instance, 1. “We have no idea that it [sic] is a copy of a necessary connection” to 2. “We have no idea that is an idea of a necessary connection”’ (p. 29; my emphasis, except for ‘of’). Landy says that this conclusion is unacceptable and that Kant will block it by denying the Representational Copy Principle. A second feature of Hume’s account is an account of how we represent a state of affairs consisting of a complex of objects. According to Landy, this account ‘is captured in the Humean slogan that a representation of a complex is just a complex of representations’ (p. 44). This account does not work, says Landy, because it is incapable of representing the specific aspect of the representational picture which is relevant to the representation. Help arrives in the form of Kant’s doctrine that representations of complexes are unified intuitions created by the application of concepts to the (otherwise unstructured) elements of an intuition, where concepts are understood as inference rules. It is the incorporation of these rules into the Kantian representation that provides the means of specifying aspect-relevance that the Humean representation cannot.
Landy’s first move in giving an account of Kant’s theory of the representation of complex states of affairs is to claim that the manifold of intuition (the elements that are contained in an intuition) that arrives via sensibility arrives unstructured and that whatever structure it receives is provided by the Understanding in its capacity to ‘combine’ intuitions into a unified representational whole.Footnote 4 There are numerous texts in which Kant affirms this, for example, at the beginnings of both the A and the B Deductions, at A98–9 and B129–30, respectively. How do we create unity-of-intuition out of these elements?
The way that we pull the representations of the parts of a triangle (here its three sides) into a complex representation of a triangle is by relating these representations to one another via a rule for constructing triangles … A concept-qua-inferential-rule does exactly this. It determines what relations must hold between representations … the concept triangle would license the inference from ‘This line segment is part of a triangle’ to ‘There are line segments that intersect each of the ends of this line segment that also intersect each other.’ It is such rules, Kant is saying, that necessarily guide the formation of a complex representation of a complex object as complex. (p. 140)
Any account of Kant’s doctrine of unified intuitions has to cope with two main questions. (1) How are unified intuitions different from judgements, given that the same function is at work in both (A79/B105)? Landy’s answer to this question is indicated in the words ‘by relating these representations to one another via a rule’ (my emphasis). The ‘representations’ mentioned in the quotation are the elements given in intuitions and Landy will argue that these just are sensations. So Landy’s claim is that sensations are related to one another by rules of inference in unified intuitions, propositions are related to one another by rules of inferences in judgements. This is how unified intuitions and judgements differ. How this is possible and how it works in detail are the subjects of the second half of Chapter 3. His discussion there is ingenious and novel and makes an important contribution to the literature, but I will not have space to go into details here. The other question is this. (2) What is the order of precedence between judgements and unified intuitions in Kant’s cognitive science? Landy’s answer is that judgements that we make when perceiving things impose an inferential structure on sensations that is analogous to the inferential structure in the judgement (p. 148). The word ‘analogous’ is important here since inferential relations hold in the proper sense only between propositions, and sensations are not propositions.
Now I turn to Landy’s account of ‘the unity of the proposition’ in Kant. In Kant’s language this is the question of the unity of various forms of judgement. The general answer is that each of the twelve forms of judgement are given their characteristic unity by the corresponding functions of the understanding, the Categories. (‘The functions of the understanding can therefore all be found together if one can exhaustively exhibit the functions of unity’, A69/B94.)
Focusing on subject-predicate judgements, here is a key text Landy relies on: ‘In every judgement, accordingly, there are two predicates that we compare with one another, of which one, which comprises the given cognition of the object, is the logical subject, and the other, which is to be compared to the first is called the logical predicate’.Footnote 5 How are these two things related to one another? Starting with an analytic proposition, Landy says:
Notice that what ‘a body is divisible’ means is that if one thinks of something using the concept ‘body,’ one thereby also thinks of it using the concept ‘divisibility.’ Part of what it is to be the concept ‘body’ is to play a role in inference whereby one can draw the inference from ‘this is a body’ to ‘this is divisible.’ (p. 68)
Obviously the kind of inference involved here is a logical inference. Landy will later argue that these inferences are immediate inferences (‘material inferences’ in Kant’s terminology), not syllogistic enthymemes, and in the case of sortal concepts (e.g. elephant) these inferences will be counterfactual inferences. What grounds the inferences is normativity, a commitment to draw the inferences that has the force of normative necessity. Landy maintains that Kantian necessity, including nomic necessity, derives from the normative necessity to draw inferences.Footnote 6 Landy now claims that this is ‘not limited to analytic judgments but extends to synthetic ones as well’, presenting another example from Kant: ‘An example of a synthetic proposition is, to everything x, to which the concept of body (a + b) belongs, also belongs attraction’ (Logic, 9: 111) (p. 68). In such a case the inference would be from ‘This is a body’ to ‘This exerts a force of attraction’, where the inference warrant is nomic rather than logical, based as just indicated on normative necessity.
We now can return to the problem of the unity of judgements. We know that Kant’s general account is that the functions of the understanding afford the unity of judgements, but we do not yet know how. Landy quotes (p. 94) the following passage as providing the key: ‘By a function, however, I understand the unity of the action of ordering different representations under a common one’ (A68/B93). Again, this seems to best fit the subject-predicate form of judgement (categorical judgements), so we will restrict our consideration to those. Landy explains Kant’s meaning: ‘Concepts rest on functions, on the unity of the action of ordering different representations under a common one’ (p. 94). He previously had considered the possibility that, for Kant, this function brings particular-entities under universal-entities,Footnote 7 but rejects it in favour of an explanation in terms of inferences: ‘We represent “x” as larger than “y” by licensing, forbidding etc., certain inferences … “x” and “y” thus become ordered under the common representation “larger than”’ (p. 94). So subsuming a representation under a common concept is what gives unity to subject-predicate judgements and committing ourselves to draw inferences from the subject to the predicate counts as subsuming the subject under a common concept because the same inferences can be applied in general to many other cases. This is Landy’s account of the unity of the proposition in Kant.
Non-Conceptualism Rises to the Challenge
I now respond to Landy’s challenge by arguing that taking concepts as rules for constructing images gives us a better understanding both of Kant’s doctrine of the unity of the proposition and of his doctrine of the constraint on judging provided by empirical intuitions. Because of space limitations and because I have made parts of the case elsewhere,Footnote 8 the argument here will be a sketch, somewhat dogmatically presented. The argument makes two assumptions. The first is that sensations constitute a field of elements with a spatial arrangement imposed by the form of outer intuition given prior to conceptualization. In virtue of this there is a class of empirical objects that are given in sensibility and directly represented by empirical intuitions that are undetermined by the categories. These are the ‘appearances’ of A20/B34.Footnote 9 The second assumption is that Kant’s transcendental idealism is the doctrine that what we take to be ordinary physical objects are intentional objects created partly as a projection of aspects of our sensibility and partly as a result of image-constructing and image-regulating rules of the understanding.Footnote 10
Re: The Unity of the Proposition
On Landy’s account the unity of a subject-predicate proposition like ‘This body is divisible’ consists of an inferential link between propositions formed from the subject and the predicate, namely, ‘This is a body’ and ‘This is divisible’. Now the latter two ‘base propositions’, as I will call them, are themselves in need of an account of unity. Either it is the same account or it is not. If it is the same then there have to be two more basic base propositions for each base proposition, and so on, ad infinitum. This is unacceptable, so the account of the unity of the base propositions will have to be different, one not involving inferences. As noted above, Landy takes the question of the unity of the proposition for Kant to be the question how intuitions are subsumed under concepts. I contend that this is just what the account of concepts as image-constructing rules affords. (Details follow in the next section.) For confirmation, consider the following quotation from the chapter on the Schematism. (Bear in mind that in this chapter the geometrical concept of a circle is a rule for constructing an image of a circle.)
the empirical concept of a plate has homogeneity with the pure geometrical concept of a circle, for the roundness that is thought in the former can be intuited in the latter. How is the subsumption of the latter [intuitions] under the former [pure concepts of the understanding], thus the application of the category to appearances, possible since none would say that the category, e.g. of causality, could also be intuited through the senses …? (A137–8/B177–8)Footnote 11
The emphases are Kant’s and he implies here that the subsumption of intuitions under concepts occurs paradigmatically in the case where a geometrical image-constructing rule is somehow applied to an intuition. This poses a problem for the application of categorial concepts to intuitions. It is as a solution to this problem that the doctrine of the schematized categories is proposed by Kant. According to that doctrine the schematized categories of quantity and quality are rules that construct determinate intuitions from indeterminate ones, the others are rules that regulate and qualify those constructions.
Re: The Empirical Constraint
On Landy’s account, there is no arrangement among sensations that is given, hence there is no empirical constraint on the arrangement of our representations of arrangements of objects that is given. The constraint occurs, rather, because the sensations are constituents of the unified intuitions that represent the objective arrangements (p. 159). But what need to be responsive to empirical constraints are the properties of objects, what they are represented as being, so what is needed in the way of empirical constraints is not provided on Landy’s account. I maintain that this can be fixed only by denying the conceptualist’s main thesis: that sensations do not come already arranged in spatial form. Denying this is our first main assumption. We can now explain the empirical constraint if we suppose that the imagination, in ‘apprehending’ such a structured set of sensations (an empirical intuition of an appearance in the sense of A20/B38) as a spatial form, traces the outline of the intuition by means of one or more rules of the productive imagination for pure geometrical concepts. I suggest that this is what Kant means when he says that the ‘very same formative synthesis by means of which we construct a figure in imagination is entirely identical with that which we exercise in the apprehension of an appearance in order to make a concept of the experience of it’ (A223/B271). Succeeding at this task constitutes subsuming the intuition under a common representation because it is in the nature of rules that the same rule can be applied to many items – intuitions in this case. Returning now to the issue of the unity of the proposition, we can see that the unity of the most basic of the base propositions can be explained as a subsumption of conceptually unstructured empirical intuitions in the sense just explained.Footnote 12