Perhaps the most famous and widely quoted, but I think also the most generally misunderstood, line in Kant’s Critique of Pure Reason is this pithy slogan: "Thoughts without content are empty, intuitions without concepts are blind" (CPR A51/B76). Leaving aside empty thoughts, is Kant saying that intuitions without concepts simply do not exist or are meaningless? Or is he saying that intuitions without concepts do exist and are meaningful, but in a way that is sharply different from that of concepts? My aim in this paper is to relate Kant’s distinction between intuitions and concepts to the contemporary debate about nonconceptual mental content. I will argue that Kant not only defends the existence and meaningfulness of nonconceptual content, but also offers a fundamental explanation of nonconceptual content that can be directly transferred to the contemporary debate about nonconceptual content and significantly advance that debate.

For the purposes of this paper, "cognitive content" is mental representational content, whether object-directed (intentionality) or self-directed (reflexivity); and for every type of cognitive content there is a corresponding cognitive capacity by means of which a creature generates or grasps, possesses, and deploys that content. We could allow for nonconscious cognitions. But the topic of nonconscious mind is one of those philosophical black holes of infinite density and little hope of return: so to keep things fairly simple, I will focus almost exclusively on conscious cognitions. Against this backdrop, the notion of nonconceptual content can then be contextually defined by way of a two-part thesis which says (a) that there are cognitive capacities which are not determined (or at least not fully determined) by conceptual capacities, and (b) that the cognitive capacities which outstrip conceptual capacities can be possessed by rational and non-rational animals alike, whether human or non-human. This thesis also expresses the doctrine of "nonconceptualism."

The recent explosion of philosophical interest in nonconceptual content can be traced directly back to Kant’s first Critique via the work of Gareth Evans, with assistance from José Bermúdez, Tim Crane, Fred Dretske, Richard Heck, Susan Hurley, Sean Kelly, M.G.F. Martin, Christoper Peacocke, Robert Stalnaker, Michael Tye, and others. In The Varieties of Reference, Evans argues for the existence and representational significance of nonconceptual content in direct singular reference, demonstrative sense perception, self-identification, and singular perceptual judgments. Evans’s argument depends explicitly and heavily on Russell’s notion of "knowledge by acquaintance," which in turn is an updated version of Kant’s notion of intuition (Anschauung). So the historical provenance of the contemporary notion of nonconceptual content leads directly back to Kant’s notion of intuitional content.

Indeed, the contemporary distinction between conceptual content and nonconceptual content is essentially the same as Kant’s distinction between concepts (Begriffe) and intuitions. Correspondingly, the contemporary distinction between conceptual capacities and nonconceptual capacities is essentially the same as Kant’s cognitively seminal distinction between understanding (Verstand) and sensibility (Sinnlichkeit). The understanding vs. sensibility distinction is cognitively seminal because it exhausts the "fundamental sources of the mind": the understanding is the logical, discursive, and proposition-forming capacity of the mind, which produces concepts as outputs; the sensibility is the affective, perceptual, and imaginational capacity of the mind, which produces intuitions as outputs; intuitions and concepts together "constitute the elements of all our cognition"; and there are no other fundamental cognitive faculties (CPR A50/B74). Otherwise put, Kant’s cognitively seminal distinction captures the difference between the rational or higher-level powers of the human mind, and the sub-rational or lower-level powers of the human mind. It is not to be assumed that rational animals do not also have the sub-rational or lower-level mental powers; on the contrary, for Kant all rational animals also have sub-rational or lower-level mental powers that they share with non-rational animals, whether human or non-human.

Needless to say, there are also some critics of nonconceptual content who deny both its existence and its representational significance. These include Bill Brewer, John McDowell, and Sonia Sedivy. More precisely, these critics want to assert the two-part thesis (a*) that all cognitive capacities are fully determined by conceptual capacities, and (b*) that none of the cognitive capacities of rational human animals can also be possessed by non-rational animals, whether human or non-human. This thesis expresses the doctrine of "conceptualism."

The articulation and defense of conceptualism has led to a recent and very lively debate between nonconceptualists and conceptualists. A rather ironic (and as far as I can tell, so far unnoticed) feature of this debate is that the most vigorous and thoroughgoing conceptualist--McDowell--not only frequently cites Kant in support of his view, but even takes himself to be developing Kant’s own theories, despite the fact (which McDowell does not acknowledge) that the very idea of nonconceptual content was originally formulated and quite elaborately developed by Kant. On the other hand, it is also rather ironic that contemporary nonconceptualists have not explicitly acknowledged their debt to Kant either--perhaps just because Evans did not. But in any case the double irony strongly suggests that we might be able to learn something new and informative about nonconceptual content by revisiting Kant’s original theory of it.

The recent debate about the existence and representational significance of nonconceptual content has been well-focused on modes of sense perception--including direct or nonepistemic perception, perceptual belief or judgment, perceptual memory, perceptual intentional agency, perceptual phenomenal consciousness, and perceptual self-consciousness. But three basic problems have plagued the whole discussion: (1) the lack of a suitably fine-grained classification of different types of nonconceptual content (the classification problem), (2) the lack of a corresponding account of the nature of concepts (the concept problem), and (3) the worry that there may in fact be no unitary phenomenon of nonconceptual content to be explained (the unity problem). It seems to me, however, that at least the first two of these basic problems can be, if not solved, then at least pre-emptively mitigated.

In response to the classification problem, I will say that nonconceptual content is cognitive content that either (i) lacks concepts altogether, (ii) does not require the correct application of concepts even if it happens to include concepts, (iii) does not require concepts even if it happens to include concepts that correctly apply, or else (iv) requires concepts but does not require the possession or self-conscious rational grasp of those concepts by the user of those concepts. For later purposes I will dub these four different kinds of nonconceptual content respectively very strongly nonconceptual content, fairly strongly nonconceptual content, moderately nonconceptual content, and weakly nonconceptual content. The degree of strength is in inverse proportion to the extent to which the conceptual abilities of the representing subject determine the form and content of mental representation: so the less that conceptual abilities are involved in determining representations, the greater the degree of nonconceptuality in those representations.

In response to the concept problem, I will say that concepts are (1) abstract semantic items, that is, fine-grained decomposable meanings that correspond on the one hand to linguistic predicates and other general terms, and correspond on the other hand to properties (i.e., rough-grained intensions, whether intrinsic or extrinsic to objects, and whether relational or non-relational), and also (2) psychological items in the triple sense that they are (a) tokened in some particular conscious mental states, (b) express subjective modes of presentation in affect or emotion, perception, judgment, thought, and intentional action, and (c) entail the existence of psychological capacities for generating or grasping, possessing, and applying concepts. More generally, concepts are at once the basic objects of conceptual analysis, psychological rules for classifying and identifying perceptual objects, and the basic elements of cognitive rationality. Finally, concepts also satisfy Evans’s "generality constraint," to the effect that the subsumption of an object under a concept implies a dual pair of cognitive capacities for applying that same concept to distinct objects and for applying different concepts to the same object.

Assuming that these pre-emptive mitigations of the classification and concept problems are at least prima facie plausible, that reduces our task load and leaves us with the unity problem. In section III, I will show that Kant gives various arguments for the existence and representational significance of very strongly, fairly strongly, moderately, and weakly nonconceptual content in inner sense and outer sense (innere Sinn, äussere Sinn), feeling or affect (Gefühl), imagination (Einbildungskraft), sense perception (Wahrnehmung), and judgment (Urteilskraft). Even more importantly however, as I will show in section IV, in the Transcendental Aesthetic section of the first Critique he traces back the very possibility of nonconceptual content to the representations of space and time, which in turn are necessary and non-empirical or a priori conditions of every mental representation generated by sensibility. These are what Kant calls "the forms of intuition." Kant also famously claims in the Transcendental Aesthetic that we can have direct nonconceptual representations of the forms of intuition as unique non-empirical objects, and he calls these representations "pure intuitions" and sometimes "formal intuitions." Pure or formal intuition in turn is taken by Kant to be the semantic and epistemic foundation of mathematics, or more precisely, of arithmetic and geometry. Elsewhere I have tried to show how Kant’s strange-sounding claim that the pure intuition of time is the semantic and epistemic foundation of arithmetic can make sense and even be defensible. But as I say, what I want to highlight in this paper are the forms of intuition. If Kant is right, then there are two and only two forms of intuition, namely, the a priori representations of space and time; and the representations of space and time are not only presupposed by all nonconceptual content but also account for the existence, cognitive significance ("objective validity"), and psychological coherence ("subjective validity") of every type of nonconceptual content. In other words, if Kant is right, then the forms of intuition provide a fundamental explanation of nonconceptual content. Moreover I believe that Kant’s fundamental explanation can be directly transferred to the contemporary debate about nonconceptual content, and in fact significantly extend it (section V). For if Kant is right, then all the different types of nonconceptual content have the same underlying nature--namely, that they are all grounded on intrinsic phenomenal spatiotemporal structure, and thereby necessarily include a spatiotemporal phenomenal field as a representational platform for the other types of cognitive content--and this solves the unity problem about nonconceptual content.

According to Kant, the central fact about the mind is its capacity to represent (vorstellen), i.e., that it has something "to put before" (stellen … vor) it: and this something is a mental "representation" (Vorstellung) (CPR A320/B376-377). Representations can be either conscious or nonconscious (CPR A78/B103). Consciousness (Bewußtsein) adds subjective unity, or an egocentric organization, to a representation (CPR B139). In turn, every conscious representation has both (i) a "form" (Form) and (ii) a "matter" (Materie) or "content" (Inhalt) (CPR A6/B9) (JL Ak. ix. 33). Materie is sensory content (more on this in the next paragraph). Inhalt by contrast is intensional content or cognitive-semantic content: what Kant calls its "sense" or Sinn, and "meaning" or Bedeutung (CPR A239-240/B298-299). The form of a conscious representation in Kant’s sense is somewhat similar to what Descartes called the "formal reality" of an idea, and the intensional content of a conscious representation in Kant’s sense is somewhat similar to what Descartes called the "objective reality" of an idea. More precisely, for Kant the form of a conscious representation is its phenomenal character. Phenomenal characters of a conscious representation, in turn, can be either figural or qualitative. Figural phenomenal characters are spatiotemporal. Qualitative phenomenal characters include (a) the difference between clarity and unclarity, and between distinctness and indistinctness, (b) different subjective attitudes of all sorts, or what Locke called "postures of the mind," including but not restricted to propositional attitudes, and (c) and our direct conscious awareness of and ability to distinguish between and generalize over types of mental acts or mental operations of all different sorts (e.g., analysis, synthesis, memory, imagination, thought, judgment, etc.), which Kant calls "reflection" (Überlegung) (CPR A260/B316) and which is somewhat similar to Locke’s "ideas of reflection." By contrast, the intensional content of a conscious representation is what it is about, or its topic. The intensional content of a conscious representation can be held fixed while varying its qualitative phenomenal character (say, from unclearly seeing A to seeing A clearly; or from asserting that P to doubting that P to denying that P); and the qualitative phenomenal character can also be held fixed while varying its intensional content (say, from being a memory of A to being a memory of B). But an individual representation is uniquely determined by its intensional content and not by its qualitative phenomenal character.

Conscious representations can be either subjective or objective, but in either case are necessarily accompanied by sensations (Empfindungen). Sensations--or what we would now call "qualia"--are intrinsic non-relational phenomenal features of all conscious representations. More precisely, sensation is "the effect of an object on the capacity for representation, insofar as we are affected by it" (CPR A19-20/B34), or in other words, sensation is the subject’s direct response to endogenously- or exogenously- caused changes in its own state. Endogenously-caused sensations are "subjective sensations," or feelings (CJ Ak. v. 206), and exogenously-caused sensations are "objective sensations," such as the sensations that accompany the perception of external objects (CJ Ak. v. 206). Whether subjective or objective however, sensations are always cognitively transparent features of the mental states in which conscious representations occur, in the sense that they interpose no intensional--i.e., cognitive-semantic--content between the conscious subject and its representations. This is because sensations refer only to the conscious subject’s direct response to changes in its inner or outer world: "a perception (Perception) that refers to the subject as a modification of its state is a sensation (sensatio)" (CPR A320/B376).

Because sensations are cognitively transparent, they must be distinguished from both subjective conscious representations and objective conscious representations alike. Subjective conscious representations are conscious awarenesses of "mere appearances" (bloße Erscheinungen) (CPR A46/B63), or the flotsam and jetsam of representational life, such as the phenomenal mental images (Bilder) that are constantly generated in the course of conscious psychological processes by the empirical imagination (CPR A141/B181), but may or may not have any coherence or representational significance. So in other words, a subjective conscious representation is a loosely-organized and relatively unstructured conscious state, the mere result of what Hume called "the association of ideas," and what Kant in the A edition of the first Critique calls the "empirical synthesis of reproduction" (CPR A101). By sharp contrast however, an objective conscious representation, or cognition (Erkenntnis), is always either outwardly directed to some object or another and thereby has what we would now call "intentionality," or else it is self-directed and reflexive. Self-directed and reflexive cognition for Kant is thus "meta-cognition," or objective conscious representation of the subject of conscious representation. Now cognitions--conscious mental states with intentionality or reflexivity--are of two distinct kinds: (1) intuitions, and (2) concepts (CPR A320/B376-377). So far, so good. But here is where things get fairly tricky.

And that is because Kant defines intuitions and concepts in such a way that they are logically independent of one another; yet he also explicitly asserts that they are cognitively complementary and semantically interdependent. This is what I will call the togetherness (of intuitions and concepts) principle :

These texts have led many readers and interpreters of Kant--e.g., McDowell--to deny the cognitive and semantic independence of intuitions and concepts. But I think that this interpretation of the togetherness principle, despite its being widely held, is wrong. It is wrong not only because it does not conform to what Kant actually says, but also because it pays insufficient attention to the fine-grained details of Kant’s cognitive semantics. As I have argued in detail elsewhere and will not repeat here, what Kant is actually saying in the famous texts at A50-51/B74-76 is that intuitions and concepts are cognitively complementary and semantically interdependent for the specific purpose of constituting objectively valid or empirically meaningful judgments. But from this it does not follow that there cannot be "empty" concepts or "blind" intuitions outside the special context of empirically meaningful judgments. ‘Empty concept’ for Kant does not mean meaningless concept: rather it means concept that is not empirically meaningful, and there can be very different sorts of concepts that are not empirically meaningful. Some concepts that are not empirically meaningful are indeed meaningless (or at least necessarily uninstantiated) in the sense of being either nonsensical or conceptually absurd, e.g., the concept of a furiously-sleeping green idea or of a round square. But for Kant there can also be concepts that are not empirically meaningful yet still fully intelligible, for example, concepts of things-in-themselves or noumena (CPR B148-149, A238/B293, B307). Similarly, ‘blind intuition’ for Kant does not mean meaningless intuition: rather it means nonconceptual intuition, and as I suggested above there can be at least four different ways in which a cognitive content can be nonconceptual, hence four different ways in which an intuition can be nonconceptual--namely, very strongly, fairly strongly, moderately, and weakly. So Kant’s term-of-art ‘blind intuition’ no more implies the denial of intuitional cognition, than our contemporary psychological term-of-art ‘blindsight’ implies the denial of visual cognition.

So my first basic point about intuitions and concepts for Kant is that despite its being true, according to the togetherness principle, that they must be combined with one another in order to generate empirically meaningful judgments, nevertheless they can also occur independently of one another and still remain meaningful. And in particular, to the extent that intuitions are logically independent of concepts, they are nonconceptual mental contents. Now I need to say more about the nature of an intuition.

Intuitions for Kant are objective cognitions that are (i) immediate, (ii) sense-related, (iii) singular, (iv) object-dependent, and (v) prior to thought. Here again I have argued in detail for this interpretation elsewhere, so will also not repeat that argumentation here. The two important things for our present purposes are that these five features are individually necessary and jointly sufficient for any objective cognition’s being an intuition, and that the fifth feature is the same as the nonconceptuality of an intuition. I will very briefly explicate the first four features and then focus on the fifth feature.

(i) Kant says that an intuition "refers immediately (bezieht sich unmittelbar) to the object" (CPR A320/B377) and again more explicitly that

(ii) Kant says that "our nature is constituted that our intuition can never be other than sensible" (CPR A51/B75), and again more explicitly that

(iii) Kant says that intuition "refers immediately to the object and is singular (einzeln)" (CPR A320/B377), that "an intuition is a singular representation" (JL Ak. ix. 91), and that

(iv) Kant says that "our mode of intuition is dependent on the existence (Dasein) of the object" (CPR B72) and that "an intuition is such a representation as would immediately depend upon the presence (Gegenwart) of the object" (P Ak. iv. 281). This is the object-dependence of intuition, , or otherwise put, its veridicality. In other words, intuition is essentially a relational form of cognition, in that the existence of the object of intuition is a necessary condition of both the objective validity or cognitive significance of the intuition and also the existence of the intuition itself: if the putative object of an intuition fails to exist, then it is not only not an objectively valid intuition, it is not even authentically an intuition (P Ak. iv. 282) but rather only an output of our faculty of imagination (CPR B278). By contrast, a concept can still both exist and be objectively valid even if it is not satisfied by anything in the actual world, so long as it can be satisfied by something in some other possible world (CPR A239/B298).

(v) This brings us finally to an intuition’s priority-to-thought. Here Kant says that "that representation that can be given prior to all thinking is called intuition" (CPR B132). All thoughts involve concepts. Furthermore it is clear that this priority of intuition to thought is both temporal and logical. Thus an intuition can actually occur without any corresponding concept, and also logically independently of any concept:

Now for some examples and elaboration. Kant explicitly asserts the existence and representational significance of nonconceptual intuitional content in inner sense and outer sense, feeling or affect, imagination, perception, and perceptual judgment.

Inner sense is a temporally successive stream of phenomenal mental contents or states in time, by means of which a conscious subject directly intuits herself:

In rational animals like us, inner sense is always accompanied by a capacity for "apperception" or self-consciousness (Selbstbewußtsein), which is a capacity for forming self-directed judgments, and thereby imposing a higher-order unity on all the cognitive faculties and their representational outputs. This capacity for self-consciousness--which Kant calls "transcendental apperception"-- is necessary for the representation of determinate objects and also constitutes an "objective unity" of consciousness (CPR B136-B143), by virtue of its introducing conceptual and propositional logical form into the structure of every representation that is accessible to self-consciousness. In this way, the capacity for self-consciousness necessarily implies conceptual abilities, and this necessary connection is captured by the characteristic self-directed discursive representation "I think" (CPR B131-132). Conversely all conceptual abilities have the capacity for self-consciousness as a necessary condition (CPR B133-134 n.). But at the same time Kant holds that it is possible for non-rational animals--and in particular human infants and some non-human animals--to have inner sense without apperception (A Ak. vii. 127-128) (PC Ak. xi. 52), hence consciousness without self-consciousness. Indeed Kant explicitly notes that his contrast between inner sense and apperception sets his philosophical psychology sharply apart from earlier systems: "it is customary in the systems of psychology to treat inner sense as the same as the faculty of apperception (which we carefully distinguish)" (CPR B153). Therefore it is also possible for non-rational animals to have both intuitions in inner sense and also consciousness without any concepts, which directly implies the very strong nonconceptuality of intuition in inner sense and of consciousness.

Outer sense is a cognitive capacity for representing objects outside the subject in space: "by means of outer sense … we represent to ourselves objects as outside us, and all as in space" (CPR A22/B37). Kant holds that it is possible for non-human animals—e.g., an ox--to have outer sense intuitions of material objects in space--e.g., a barn-gate--without any corresponding concepts (FS Ak. ii. 59). Even more strikingly, he also holds that it is possible for rational humans to have outer sense intuitions of material objects--e.g., a house--while lacking the concepts necessary for seeing it as a house:

Kant also offers a slightly different argument for the fairly strong nonconceptuality of intuition in outer sense. This argument is from the existence of what I will call "veridical illusions"--his apt example is the fact that the rising moon always looks larger when it is nearer the horizon than when it is higher in the sky (CPR A297/B254)--whereby a real material object is directly intuited in outer sense but the corresponding concept of that object is false of it (as in the perceptual judgment, "That moon over there just above the trees is bigger than the normal moon"). In veridical illusions we can see that "truth and illusion are not in the object, insofar as it is intuited, but in the judgment about it insofar as it is thought" (CPR A293/B350). In other words, the intuition in outer sense is veridically illusory in the sense that it truly presents an actually existing object (the moon) just as it would look to any creature equipped with our cognitive faculties under those contextual conditions (i.e., as seeming bigger near the horizon than when it is higher in the sky), so the error or illusion lies in the corresponding concept and not in outer sense, which again directly implies the fairly strong nonconceptuality of outer sense. It also directly implies the evidential character of some nonconceptual contents, since for Kant perceptual judgments that are based on veridical illusions are non-inferentially warranted even though false.

Feelings or affects also supply some very interesting examples of nonconceptual content. For Kant, feelings are subjective sensations that necessarily involve either pleasure or pain (although they need not be exhausted by their pleasure/pain component). Pleasure and pain in turn are modes of "the feeling of life" (CJ Ak. v. 204), or vital subjective experience that expresses our existence as living organisms and embodied minds. In aesthetic experience of the beautiful, according to Kant, we get a "disinterested pleasure" that expresses the harmonious and life-enhancing interaction between our various cognitive faculties--and in particular between the understanding and the imagination--as they jointly operate in order to represent the phenomenal form of the beautiful object (CJ Ak. v. 217-219). On the basis of this disinterested pleasure, we non-inferentially judge that the object--say, this rose--is beautiful. But at the same time, "the judgment of taste … determines the object, independently of concepts, with regard to satisfaction and the predicate of beauty" (CJ Ak. v. 219). In other words, even though the object falls under some concept or another (we also see it as a rose), this conceptual fact is wholly irrelevant to its being beautiful, since its being beautiful consists merely in the relation between its phenomenal form and the pleasure we experience in the harmonious interplay of our cognitive faculties. So despite the fact that the judgment of taste includes concepts and as it happens those concepts correctly apply to that object (i.e., this is indeed a rose), nevertheless even if the concepts were false of the object and even if it did not actually exist as such, still the aesthetic judgment of taste has a direct object. A rose by any other conceptual label would look just as beautiful: "I do not need [a concept] in order to find beauty in something" (CJ Ak. v. 207). This in turn directly implies the moderate nonconceptuality of feeling.

The role of the imagination in the nonconceptuality of feeling is crucial. The imagination belongs to sensibility (CPR B151), and is a cognitive function of intuition: "imagination … [is a power] of intuiting even when the object is not present" (A Ak. vii. 153). Superficially this formulation is inconsistent with Kant’s definition of intuition, since intuition includes object-dependence as a necessary condition. So it could legitimately be read as a case of Kant nodding. But more charitably, I think that the best overall construal of what he is saying is that imagination is essentially intuition minus object-dependence: so imagination is an immediate, sense-related, singular, and nonconceptual cognitive capacity that can represent either existing or non-existing objects. Or otherwise put, imagination is quasi-intuition. That there is a cognitive function for representing objects that do not exist--or at least do not presently exist--is obvious in the case of sensory reproduction (i.e., memory) and mental imagery. More generally however, the faculty of imagination for Kant is not merely a capacity for reproducing sensory representations and generating mental images, but rather an all-purpose cognitive engine for representational synthesis or mental processing (CPR A78/B103). Under the rubric of "productive imagination" (CPR B152) (A vii. 167), the imagination is also a specialized or dedicated faculty for producing a certain special sort of representations called "schemata" (CPR A137-142/B176-181). Schemata are essentially figural, or spatiotemporally patterned, because they are the direct result of the "figurative synthesis" or synthesis speciosa of the productive imagination (CPR 151). But they are also inherently sortal because they can be used to organize sensory images under concepts: "this representation of a general procedure of the imagination for providing a concept with its image is what I call the schema for this concept" (CPR A140/179-180). More precisely, and translated out of Kant-speak for a moment, schemata can directly encode both sensory and discursive information in a phenomenal spatiotemporal structural format--Kant’s example is a monogram (CPR A142/B181), but a better example would be a map--and thus are mental icons, outlines, models, or templates of what they represent. I will return in sections IV and V to the role of spatiotemporal representation in nonconceptual content. But for the moment we need only note that insofar as schemata are cognitive functions of intuition or quasi-intuitions, and thereby not inherently conceptual in nature, it follows that the content of imaginational representation is nonconceptual.

We have now seen how Kant is committed to the existence of very strongly, fairly strongly, and moderately nonconceptual content in inner sense, outer sense, feeling or affect, judgment (i.e., the judgment of taste), and imagination. And I have also highlighted the role of spatiotemporal representation in nonconceptual content for later consideration. But there are some further implications of Kant’s arguments that we should also briefly note. Since a sense perception of an object is simply a conscious outer intuition of a material object in space (CPR B160, B275-279), it follows directly from the very strong and fairly strong nonconceptuality of outer sense that sense perception is also very strongly and fairly strongly nonconceptual. Moreover, consider a perceptual judgment like "This bent stick in the water is three feet long," accompanied by a visual sense perception that provides good prima facie evidence for the truth of that judgment. Suppose however that what you are actually looking at is a (relatively) straight snake in a pond filled with gin and it is actually only two feet long. This veridical illusion guarantees the nonconceptuality of that perceptual judgment, so it follows that for Kant perceptual judgments are also fairly strongly nonconceptual. Fairly strong nonconceptuality may be inconsistent with true perceptual judgments, but it is perfectly consistent with false ones.

Now if sense perceptions and perceptual judgments alike are fairly strongly nonconceptual, then so are both perception-based desires and volitional intentions. This becomes obvious when we consider that I can want that bent stick in water (for my private collection of extremely interesting bent sticks, of course) and also intend to grab that bent stick in water (so that I can proudly take it home with me). More generally, to the extent that desires and volitional intentions are all based on appearances of the good, that is, on things that seem good for me, it is obvious that not only can I be quite wrong about whether the F that I want or the F that I intend to act upon is in fact good for me, but also more generally I can be quite wrong about whether this or that is in fact the F, but nevertheless nonconceptually want or intend to act upon precisely this or precisely that.

It should now be clear that Kant has solid reasons for holding that very strong, fairly strong, and moderately nonconceptual content are quite common and indeed pervasive in the mental lives of animals, including rational animals like us. There is however one other kind of nonconceptuality noted by Kant that we need to look at very briefly before moving on, because it is interestingly and importantly different from the other kinds.

So far I have concentrated on cases of nonconceptuality in which for one reason or another, a human or non-human animal’s capacity for sensibility in some way cognitively dominates over the capacity for understanding, even if concepts are present in mental content and even if concepts happen to be correctly applied. But Kant also points up cases in which there is a cognitive dominance of sensibility over the understanding even though the cognition in question is otherwise intrinsically conceptual in character. These are cases in which the sensibility-driven use of a concept dominates over the possession of that concept, or more precisely, cases in which a concept is correctly applied by a subject even though the subject cannot self-consciously and rationally (i.e., logically and analytically) grasp that concept. Kant says:

As theorists of nonconceptual content, what we need to know more precisely is just what sort of cognitive activity is actually going on in weak nonconceptuality. Unfortunately Kant does not tell us explicitly, except for an intriguing analogy between a subject’s conscious awareness of the intensional content of her concepts and our cognition of maps (JL Ak. ix. 64). But contemporary work in the cognitive psychology of concepts suggests that much of our ordinary concept-use is determined by our ability to match items in the world with "stereotypes" or "prototypes," which in Kantian terms would be schemata consisting of classificational patterns of linguistic and non-linguistic imagery of perceptually salient features of objects and situations. Assuming that this is what Kant has in mind, it implies that for the purposes of "common and practical use" of concepts, the human capacity of outer sense plus the schematizing function of the imagination can cognitively dominate over our self-conscious rational capacities for concept-possession, even in fully conceptual cognition. So despite its differences from the other sorts of nonconceptuality, instances of weak nonconceptuality seem to have essentially the same underlying nature as all nonconceptual content--namely, that human or non-human sensibility necessarily introduces some set of inherent phenomenal features, such that these features jointly guarantee that a cognitive content having just those features is logically independent of and irreducible to conceptual capacities. Let us turn now to Kant’s theory of the inherent phenomenal features that determine the nonconceptuality of a mental content.

In the first Critique, in the first section of the Transcendental Deduction of the Pure Concepts of the Understanding, Kant observes that "we have already traced the concepts of space and time to their sources by means of a transcendental deduction, and explained and determined their a priori objective validity" (CPR A87/B119-120). And in the Prolegomena Kant speaks of a "transcendental deduction of the concepts of space and time" (P Ak. iv. 285). I take a transcendental deduction to be a demonstration of the objective validity--the empirical meaningfulness or cognitive significance--of an a priori representation R (whether that representation be an a priori concept, an a priori intuition, or an a priori necessary proposition), by means of demonstrating that R is the presupposition of some other representation R*, which is previously assumed to be objectively valid (CPR A84-94/B116-127, A156/B195). What I want to do in this section, then, is to reconstruct Kant’s transcendental deduction of our a priori representations of space and time. In so doing, it will become evident that our a priori representations of space and time are for Kant the presuppositions of all nonconceptual content, and thereby provide the phenomenal features that jointly guarantee that a cognitive content having just those features is always logically independent of and irreducible to conceptual capacities.

Central to my reconstruction of this transcendental deduction is what I call a "cognitive-semantic," as opposed to either a metaphysical or an epistemic, reading of Kant’s transcendental idealism. On the cognitive-semantic reading, Kant’s theory of space and time is not first and foremost an investigation either into "the question of the ontological status of space and time," or into the question of how we obtain justified true beliefs about space and time, but instead into the a priori features of the "concepts" or representations of space and time. Kant’s turn away from metaphysics or epistemology towards cognitive semantics via his transcendental deduction of the representations of space and time implies, among other things, that his notorious thesis of the "transcendental ideality" of space and time--that space and time are nothing but a priori necessary subjective forms of human sensibility (CPR A28/44, A36/B52)--is not in fact the conclusion of Kant’s transcendental deduction of the representations of space and time, but instead a logically independent thesis that is detachable from that deduction. So I can leave aside the thorny issue of Kant’s idealism in my presentation of this deduction.

Unfortunately, Kant’s use of the term ‘concept’ or ‘Begriff’ in the Aesthetic (and also in the treatise on which the Aesthetic was originally based, On the Form and Principles of the Sensible and Intelligible World, or the "Inaugural Dissertation") is consistently ambiguous in one respect. He explicitly discusses the "concepts" of space and time. But an intermediate conclusion of the Aesthetic is that neither the representation of space nor the representation of time is a "discursive" representation or a "general concept": rather, both are intuitions and therefore not concepts (CPR A24-25/B39, A31/B47). So in order to be charitable to Kant and to avoid the absurdity of his arguing that the concepts of space and time are not concepts, we must take all his references to the "concepts" of space and time (with the single exception of a special case that I will mention in the next paragraph) to invoke a broad meaning of ‘Begriff’ that is essentially the same as that of the neutral term ‘Vorstellung’ or ‘representation’. This comports well with Kant’s usage of ‘Begriff’ in the pre-Critical writings and in the Reflexionen. It also makes sense of an otherwise unintelligible passage in the first Critique in which he explicitly distinguishes between "two sorts of concepts of an entirely different kind, which yet agree with each other in that they both relate to objects completely a priori manner, namely the concepts of space and time, as forms of sensibility, and the categories, as concepts of the understanding" (CPR A85/B118). So in order to avoid confusion, I will consistently use ‘representation’ where the broader sense of ‘concept’ is clearly intended by Kant.

Just to make things even more complicated however, Kant does speak in at least two places of "the general concept of spaces" (CPR A25/B39) and of a "general concept of space (which is common to a foot as well as an ell)" (CPR A25). Since these passages are juxtaposed with arguments against construing spatial representation as conceptual--and again on the charitable assumption that Kant is not simply contradicting himself--he must actually be arguing that despite the fact that the representation of space is not a concept, there can nevertheless be some sort of general concept of space. But this general concept of space will be parasitic on a more basic intuition of space, just as we might form the concept of "being socratic" on the basis of a direct acquaintance with Socrates himself.

For convenience in what follows, I will abbreviate ‘the representation of space’ as ‘r-space’ and ‘the representation of time’ as ‘r-time’, display the individual steps of the argument along with supporting texts (whether from the Transcendental Aesthetic or elsewhere in the first Critique), and also give a brief commentary on each step.

We are now in a position to see how the forms of intuition provide a fundamental explanation of nonconceptual content. Kant’s way of formulating this, as we have just seen, is that the forms of intuition are the conditions of the possibility of, or the presuppositions of, all nonconceptual content, as guaranteed by the transcendental deduction of r-space and r-time. But Kant’s thesis about forms of intuition can also be translated out of Kant-speak and into more contemporary terms. When we do so, we realize that he is making an intelligible, substantive, and plausible claim that significantly extends the recent debate about nonconceptual content, by solving the unity problem.

The key to recognizing Kant’s solution to the unity problem lies in the answer to the following question: is the underlying nature of cognitive content exhausted by its functional or its purely logico-rational components? Those who answer "yes" to this question will deny either the existence or at least the representational significance of nonconceptual content, whereas those who answer "no" will assert the existence and significance of nonconceptual content. Kant’s fundamental explanation for nonconceptual content via the forms of intuition gives us good reason to answer "no," and here is why.

As we have seen, forms of intuition introduce intrinsic phenomenal spatial or temporal structures directly into cognitive content--or otherwise put, all sensory representations of objects or of the subject itself via empirical intuition, are immanently necessarily informed by r-space or r-time. Now since r-space and r-time are intrinsic phenomenal structures, they are irreducible to phenomenal qualia or sensations. This is because qualia or sensations are intrinsic non-relational features of cognitive content, whereas r-space and r-time are fully relational (CPR B67). Moreover, since r-space and r-time are intrinsic phenomenal structures, they are irreducible to functional features of cognitive content. This is because functional features of cognitive content are extrinsic relations within content that trace causal mappings (and in a materialist representational framework, these mappings could ultimately be either behavioral, computational, otherwise mechanical, or neurobiological) from processing inputs to processing outputs. And finally, since r-space and r-time are intrinsic phenomenal structures, they are irreducible to purely logico-rational features of cognitive content. This is because the purely logico-rational features of cognitive content are non-phenomenal in character. In this way, the intrinsic phenomenal structures of r-space and r-time are not only required by but also directly organize and configure all phenomenal content. So while it is plausible to hold that all the conceptual parts of phenomenal content can be accounted for (non-materialistically, or materialistically) in either functional or logico-rational terms, nevertheless the nonconceptual spatiotemporal parts of phenomenal content necessarily resist functional or logico-rational reduction.

Interestingly, Kant’s thesis of the cognitive autonomy of nonconceptual spatiotemporal representation has also received some independent empirical confirmation, at least as far as spatial representation is concerned, in experiments involving commissurotomy patients. Commissurotomy is the surgical severing of the corpus callosum, which is the primary neural connection between the right and left hemispheres of the brain. Commissurotomy patients typically manifest some cognitive dissocation between types of information normally processed in the right hemisphere and types of information normally processed in the left hemisphere. In one particularly interesting experiment carried out by Colwyn Trevarthyn, a commissurotomy patient was instructed to carry out a left-hand task that referred to an object on the right side of the visual field, while focusing his vision on a central point between the two sides of his visual field. What happened was that as soon as the left-hand movement (controlled by the right hemisphere) started, the visual appearance of the object in the right-hand side of the visual field (controlled by the left hemisphere) disappeared, thus vividly indicating a strong dissociation between information processed in the two hemispheres. But most importantly for our purposes, even though the visual representation of the object disappeared, the right-hand phenomenal visual field remained both intact and continuous with the left-hand side of the visual field. In an elaboration of these results, Trevarthyn proposed "that neo[i.e., recent]commissurotomy in man may … divide cortical vision for perception of detail and identification of objects, without producing a similar division in the perception of ambient space." In other words, it is possible for commissurotomized humans to dissociate from the conceptual content of visual experiences and judgments, while still retaining the uncompromised nonconceptual grasp of visual space. So the visual representation of space appears to be both nonconceptual and also more cognitively basic than visual conceptualization.

How, more precisely, do the forms of intuition according to Kant play their constitutive role in nonconceptual content? The crucial fact is that they introduce a single spatiotemporal phenomenal framework, or a phenomenal field, into cognitive content. This phenomenal field is not merely a set of spatial and temporal relations, but also has what I will call a "designated" structure. This means that the set of spatial relations and the set of temporal relations found in the sensible experience of rational animals like us have further special constraints on them, that cannot be found in every logically possible set of spatial or temporal relations.

For example, according to Kant, the spatial part of the spatiotemporal phenomenal field is not only represented as filled with points, as figural, and as extended, but also as an oriented three-dimensional Euclidean manifold. Obviously the Euclidean designation sets it apart from non-Euclidean spaces; and just as obviously the 3-D designation sets it apart from higher-dimensional Euclidean or non-Euclidean spaces. But the oriented designation is particularly important. As Kant argues in his fascinating pre-Critical essay, "Concerning the Ultimate Ground of the Differentiation of Directions in Space," and again in his equally fascinating late Critical essay, "What is Orientation in Thinking?," all orientable three-dimensional spaces represented by creatures like us necessarily have "centered" or egocentric axes for right-left, front-back, and up-down directionality (DS Ak. ii. 381-383) (OT Ak. viii. 134-135). So according to Kant it is possible for me to cognize the difference between my right and left hands nonconceptually, merely by possessing an outer sense (P Ak. iv. 285-286).

In turn, according to Kant, the temporal part of the spatiotemporal phenomenal field is not only represented as successive, filled with moments, and as linear or one-dimensionally extended (CPR A33/B50), but it is also represented as asymmetric or irreversible in its succession (CPR A191-193/B236-238). The asymmetric character of r-time--"time’s arrow"--is of course crucial in our representation of the causal order of nature, and guarantees that our representation of material causal processes--in particular, those involving motion (CPR B48-49)--will represent such processes as always and only flowing successively forwards into the future and never backwards into the past. And obviously, as science fiction fantasies about time-reversal or time-travel show, it is not the case that every logically possible dynamic world will only flow forwards. Thus for Kant, just like the oriented directionality of space which depends solely on the constitution of outer sense and is thereby nonconceptually cognizable, so too the irreversibility of time is an essentially "centered" or egocentric feature of time that depends solely on the constitution of inner sense and is nonconceptually cognizable.

The crucial Kantian conclusion to draw from all of this, from a contemporary perspective, is that due to the fact that the forms of intuition introduce designated intrinsic phenomenal spatial or temporal structures into all human or non-human sensibility, they ground all nonconceptual content and also provide a representational platform for all the other sorts of cognitive content, and in particular those involving conceptualization or judgment. And this solves the unity problem.

If I am right, then the Kantian forms of intuition explain why nonconceptual cognitive content exists and is representationally significant. But I think that there are also two further deep lessons about the nature of cognitive content that contemporary theorists of cognition and mental content can learn from Kant’s theory of nonconceptual content.

The first deep Kantian lesson is not to assume that all features of cognitive content are either (1) intrinsic and non-relational features of content (e.g., sensory qualia) or (2) extrinsic and relational features of content (e.g., functional features), with no third option. This is because there can also be intrinsic structural features of phenomenal content, namely, spatial and temporal features of it. This in turn raises the possibility that non-phenomenal or purely rational cognitive content might also have intrinsic structural spatiotemporal features.

And the second Kantian lesson is to recognize that the "centeredness" or egocentricity of conscious cognitive content is essentially bound up with spatial orientation and temporal asymmetry. In other words, the formal intuitional spatiotemporal structure of conscious cognitive content just is its subjective or "first person" character, when content is nonconceptual. Therefore Kant is in effect telling us that in light of the cognitively basic nature of nonconceptual content, it is an animal’s spatiotemporal perspective or "unique point of view" that is basic to the subjectivity of its experience, not "the unity of consciousness," in the Kantian sense of a necessarily concept-related capacity for the self-conscious propositional unification of a phenomenal manifold of content. As commissurotomy cases vividly show, the unity of consciousness in this sense is a relatively sophisticated and relatively fragile achievement of rational animals, but unnecessary for animal consciousness and conscious animal cognition in general, whether the animal is rational or non-rational, and whether the animal is human or non-human.