1. Three kinds of blind intuition

It has been much debated recently whether there can be non-conceptual content in Kant's scheme of things. One of his most-quoted lines is ‘intuitions without concepts are blind’, which may seem to rule out unconceptualized perceptions straight away. But that is only a slogan, we need to study its context:

This does not say that intuitions cannot exist without concepts, only that they would be ‘blind’, nor that thoughts cannot lack ‘content’, only that they would be ‘empty’. A thought in Kant's sense (and Frege's) is a proposition, something true or false, a judgable, believable, assertable content, so a thought without propositional content would be a contradiction in terms. But the point about cooperation between sensibility and understanding suggests that Kant was thinking of perceptual content here. We have plenty of thoughts without present perceptual content, e.g. ‘I met Maria yesterday’, ‘I'll see her again tomorrow’, ‘Caesar crossed the Rubicon’, ‘The earth revolves around the sun’. But we could not formulate such thoughts unless we were embodied creatures who can make present-tense perceptual judgments, and that implies an indirect dependence of all our thought on sensibility.

I am more concerned here with the second claim, that intuitions without concepts are blind. Later on, at A90/B122 Kant declared that ‘appearances can certainly be given in intuition without functions of the understanding’, which suggests that there is room for blind intuitions in his philosophy (see also A111, A124, B132). But let us first clarify his notion of intuition (Anschauung):

So Kantian intuitions are mental states that relate immediately, by way of singular reference, to mind-independent objects and states of affairs when presented by causally impinging on our sense-organs.Footnote ⁶

So what can count as blind intuition? If Kant meant unconceptualized, it would be trivially analytic that intuitions without concepts are blind. A more plausible interpretation is that ‘blind’ means unconscious, and therefore not available for judgment or inference. Some will say that perception in animals is unconscious, but that depends on how we interpret the notion of consciousness. Some may say we can never know about an animal's state of mind, though we tend to be more generous to our babies. Others may reply that though animals manifestly perceive objects, they don't perceive facts, they can't see that anything is the case. But common usage and ethology agree in saying that many animals can see that a predator is approaching or a prey is escaping in a certain direction. Some philosophers and psychologists say that animals and infants apply concepts of a primitive sort,Footnote ⁷ but Kant (and Frege) had a stronger sense of Begriff according to which concepts are ‘predicates of a possible judgment’ (A69/B94), and judgments are expressible in sentences.

Are there unconceptualized perceptions in adult humans? In the Jäsche Logic Kant offered a disputable (and politically incorrect) example:

In a culturally converse situation, an archaeologist might dig up a mysterious artefact for which she cannot think of an intended use. But these are not completely blind intuitions, for both ‘savage’ and archaeologist can surely describe what they see as a material object of a certain size and shape. There is a more complete lack of conceptualization when we react behaviourally to a noxious smell or a sudden noise, or cringe on seeing something flying towards us. Unconceptualized perceptions can exist alongside or underneath our most conceptualized experience.

What about consciousness? This is of course a very hot potato in philosophy and psychology, and I can only handle it momentarily here. Can some animal perception be conscious? I submit that our untheoretical talk of consciousness does not settle the question: it is not as if there is some fact of the matter that is forever beyond our ken, but that we are somewhat unclear about what the question means. A stag hears the bellow of a rival and sees his threatening posture, and he is surely aware of it, for he gets ready to do battle himself – so why not say he is conscious of his rival? But there is no reason to say he is self-conscious, he certainly does not use a first-person pronoun.Footnote ⁹ Kant said that before using the word ‘I’ the child ‘merely felt himself; now he thinks himself’,Footnote ¹⁰ but about adult humans he declared:

But given the existence of some unconceptualized perceptions in even the most intellectual of us, we can hardly agree that they ‘are nothing for us and do not in the least concern us’, for they can affect our behaviour, if not our consciousness. Kant's main concern in the first Critique was with the necessary conditions of our self-conscious and conceptualized knowledge that he calls cognition, but he acknowledged that we have many mental states of which we are not conscious:

However conceptualization and availability to consciousness do not always coincide. We retain many memories and beliefs through the periods when we are not consciously thinking of them (Freud called them ‘preconscious’), but if there is any truth in Freudian theory of the unconscious or Sartre's conception of bad faith, people can have perceptions of the emotive or erotic meaning of someone's behaviour and speech that may resist being brought to consciousness, even if that meaning could be expressed in terms of the subject's concepts. In the Anthropology Kant alluded to repressed ideas, anticipating Freud's interpretations of the sexual meaning of many jokes, figures of speech and gestures:

Subliminal perception also involves conceptualization without consciousness, and it has been experimentally demonstrated that people's behaviour can be affected by very brief displays (not consciously noticed or verbally reportable) of an advertisement or an emotive word, where the meaning of the display must be understood at some level for the subliminal perception of it to have its distinctive effect.Footnote ¹³

I have already alluded to the converse phenomenon of human consciousness without conceptualization. One can see something indistinctly in fog, on the horizon, or in one's peripheral vision, without being able to apply any concept to it – maybe not even material object for it might be only a shadow, a reflection, or a gap between objects. When playing tennis we hit back oncoming balls without applying concepts of their speed, direction or spin. Many of us can recognize a tune or sing a passable rendition of it without resort to musical notation. We recognize people by their faces without being able to describe their identifying features in words, and we are hypersensitive to their facial expressions, gestures, and tones of voice: our evolution as a highly social species has equipped us with face-recognition modules and emotion-discerning capacities which operate before any conceptualization kicks in. I therefore propose that we can distinguish the following four levels of mentality, including three species of blind intuition:

1. a. Conceptualized and conscious. Much human perception falls under this heading, whenever we make perceptual judgments that are expressed or expressible in language. This level of mentality, which Kant calls ‘cognition’ (erkenntis) or experience (Erfahrung), is at the centre of his attention.

2. b. Conceptualized, but not available to consciousness. Subliminal perceptions, and perceptions whose content is repressed, belong here. These form one kind of blind intuition, which Kant calls ‘obscure’ (dunkel).Footnote ¹⁴

3. c. Unconceptualized, but conscious. For example, unidentified tastes, smells, noises, flashes, pressures and tickles: these are a second species of blind intuition in which we are aware of our own ‘sensations’ (Empfindung). There are also perceived but unconceptualized features of the external world in music, art, facial expressions, or tennis.

4. d. Unconceptualized, and not available to consciousness. The first effects of the environment on our sense-organs are the third and lowest level of blind intuition, which Kant calls ‘sensible impressions’ (sinnliche Eindrücke).

2. Sensible impressions or sensations?

On the first page of B, Kant used the term ‘sensible impressions’ for the raw sensory impacts that form the input to the processes involved in perception:Footnote ¹⁵

But his language in this area roams around a bewildering (baroque?) panoply of terminology, including: ‘a manifold in intuition’ (A97, 99), ‘representations, as modifications of the mind in intuition’ (A97), ‘the succession of impressions on one another’ (A99), ‘sense represents the appearances empirically in perception’ (A115), ‘the manifold of sensible representation (intuition)’ (A129, B129), ‘representations given in intuition’ (B133–4), ‘the manifold of a given intuition’ (B137), and ‘sensations’. Kant described the initial sensory stimulations as one kind of representation. which may suggest that they are available to consciousness – but his very capacious use of ‘representation’ (detailed at A320/B376) does not have that implication. Philosophers in the empiricist and sense-data tradition have assumed that perception is based inferentially on a foundational layer of introspectable mental items that have variously been called ‘ideas’, ‘impressions’, ‘percepts’, ‘sense-data’, ‘sensations’, ‘qualia’ or ‘phenomenal qualities’. (I am suggesting that Kant pointed the way towards a different view, though somewhat unsteadily)

Tyler Burge compliments Kant for improving on preceding philosophy by distinguishing between sensation, intuition and concept, which Burge lines up with his own distinctions between sensory registration, perception, and propositional thought.Footnote ¹⁶ His survey of perceptual psychology firmly distinguishes perceptual representation of physical objects and states of affairs from mere sensory registration and reaction. The latter is exemplified in molluscs closing up whenever a shadow passes over them: this is a defensive adaptation that reduces their chances of being eaten, but they do not form any perceptual representation of the cause of the shadow. A more subtle case of sensory registration is found in salmon who navigate from the far reaches of the ocean back to the stream in which they were born, guided only by extremely dilute traces in the seawater. Impressive as this is,Footnote ¹⁷ there is no reason to judge that salmon have any perceptual representation of the chemistry of the ocean, or of the location of their birthplace. There may even be some mere sensory registration in humans: many animals and insects emit chemicals known as pheromones which affect the behaviour of conspecifics, often to communicate availability for mating, and it is apparently an open question whether humans respond to pheromones, erotic or otherwise.Footnote ¹⁸

Perception proper, involving the formation of representations that distinguish and track material objects and states of affairs, takes place in a wide variety of animals. Archer fish aim a gobbet of spit at insects sitting on leaves above the water, allowing for the refraction of light through the surface and knocking the prey off its perch to be gobbled up for lunch. Jumping spiders navigate expertly in a tangle of twigs. Most mammals and birds perceive their mates, rivals, offspring, prey, or predators, tracking them through space and time through a variety of perspective, distance, lighting, and motion. Kant was in no doubt that animals perceive objects, in some sense.Footnote ¹⁹

Neuroscience investigates the mechanisms that make perceptual representation possible. In the case of vision, computational processes in the brain transform the patterns of light striking the retina into representations of objects of certain sizes and shapes at various distances. This involves mathematical transformations of very subtle geometric properties of the retinal stimulation, guided by innate neurophysiological programmes.Footnote ²⁰ We can recognize that, with hindsight, as part of the ‘working up’ of the raw sensible impressions that Kant talked of. But even in the most knowledgeable of humans such processes are not usually available to consciousness, and they are not propositional inferences. They are not normally affected by the will, and their functioning is demonstrated by the persistence of visual illusions such as the Müller-Lyer and Escher's drawings even when one knows they are illusions.

Much philosophical talk of sensations has been ambiguous and controversial. But Kant made available (though not as up-front as we might like) a distinction between sensations as conscious mental states, and sense-impressions which seldom or never reach consciousness. Admittedly his first mention of sensation as ‘the effect of an object on the capacity for representation, insofar as we are affected by it’ (A19-20/B33-4) seems to leave it open to understand the relevant effects as unconscious and physiological. However in his rather belated systematic classification of his mental vocabulary at A320/B376-7 he defined sensations as those conscious representations (‘perceptions’) that involve only a state of the subject, as opposed to ‘objective perceptions’ or cognition (see also his treatment of sensation at B207-8). So in Kant-speak, sensations are conscious representations that don't actually represent anything distinct from the subject.Footnote ²¹ He was prepared to count anything that contributes to cognition as a representation, a mental state in the widest possible sense.

In contemporary usage the word ‘sensation’ is most at home in talk of bodily sensations such as pains and itches, and the activities and pleasures involved in swimming, tennis, sex or the consumption of chocolate liqueurs. It is natural to say that our senses of touch, smell and taste involve sensations, since they involve bodily contact with whatever we sense, so that we can readily be aware both of our own sensations and of what is causing them (though often in an unconceptualized way when we cannot find words to describe a taste, a smell or a caress). Our sensations may be caused by external objects or internal bodily states, but that does not mean that they are of their causes in the representational sense. A pain can be caused by a pin, or by acid or a tumour, but it does not represent any of those things: one can be vividly aware of a pain while knowing nothing of its cause. It is not quite so natural to talk of sensations in our distance senses of hearing and sight, but Kant offered an example:

Tinnitus, in which someone is bothered by illusory ringing in the ears, would be another case. For cases of visual sensation we can think of after images, of the ‘rays’ one ‘sees’ if one looks at a light-bulb and screws up one's eyes, the redness one experiences through closed eyelids in sunlight, the shimmering effect of some subtly-patterned geometric paintings, or the double image of something too close to one's eyes. In these cases one is aware of a state of one's own visual system rather than (or as well as) something distinct from oneself. But these examples are of the illusory, the abnormal, the painful, even the medical. Are there aural and visual sensations in every normal case of hearing and sight? In most cases we simply hear or see things without being aware of sensations in our ears or eyes – we do not ‘feel our own state’, as Kant put it, though we can pay introspective attention to how things seem to us. But sensible impressions, understood as the physical effects of the stimulations of our sense-organs, are involved, by definition, in all the senses all the time.

So, contrary to the tradition of Lockean ideas, Humean impressions, Russellian/Moorean sense-data or modern-day phenomenal qualia,Footnote ²³ we should not expect to find a lower level of propositional judgments about the contents of our conscious sensory states to serve systematically as premises to justify our perceptual judgments about the external world.Footnote ²⁴ There are causal processes in perception, which psychology has now investigated in intricate detail in many creatures, but in humans these causal intermediaries are not inferential intermediaries.Footnote ²⁵ True, the more sophisticated of us can, with some mental effort, describe how things look or sound to us from where we are at particular moments of time.Footnote ²⁶ Such descriptions use in a qualified way some of our normal vocabulary for public objects and states of affairs. It requires special skill to draw in correct perspective, and to decide what colours grass looks in sunlight and shadows (and to represent that appearance in paint), though most of us can recognize that something looks spherical or heavy, or that someone looks elderly or angry. We are not restricted to the two-dimensional patches of colour beloved of the sense-data tradition, which now appears to be a mongrel offspring of two quite distinct conceptual schemes: the causal-scientific-physiological and the introspective-conscious-inferential. I propose that we accept Kant's distinction between sense-impressions, sensations, and empirical intuitions, and enforce them more consistently than he did.

3. Synthesis: psychology or philosophy?

Having firmly distinguished between intuitions and concepts, and less clearly between sensible impressions and sensations, Kant introduced his notion of ‘synthesis’:

And at this stage he suddenly introduced a third mental faculty, operating somewhere in between sensibility and understanding, called ‘the imagination’ (Einbildungskraft), to perform the function (or functions?) of synthesis:

He thus recognized that most of the mental processing in perception proceeds below our conscious awareness. (Indeed it happens very fast, for we make many of our perceptual judgments almost instantaneously) We might expect him to hand over to empirical psychology and physiology at this point, yet he went on to theorize in some detail about synthesis, even though he of all people was vividly aware of the distinction between empirical science and a priori philosophy. In the closing pages of the first Critique he declared that ‘empirical psychology must be entirely banned from metaphysics’, yet he conceded ‘a little place’ for it as ‘a long-accepted foreigner, to whom one grants refuge for a while until it can establish its own domicile in a complete anthropology’ (A848-9/B876-7). No doubt that reflected the state of psychology in 1781, but it has now progressed far beyond that, and has firmly established its proper ‘domicile’ among the natural sciences, so I submit that our understanding of perception should take account of this. Kant offered a threefold theory of synthesis in the first edition, yet in the Preface he suggested it was not essential to his argument:

The latter question sounds indeed like one for empirical psychology to answer, but Kant reiterated here that the main purpose of his Transcendental Deduction was to show that the categories must apply in our experience, as an a priori truth. That is the objective side of his argument, as explained in the introductory section common to both editions:

However in the following passage he rather surprisingly declared:

That seems to allow as a possibility precisely what the Transcendental Deduction is supposed to rule out. Some scholars used to suggest a patchwork thesis, that in the rush to publication Kant stitched together bits of draft text without sufficient care; but a more sympathetic interpretation is that it only seems at first glance that there is a possibility of uncategorized appearances, whereas the Deduction's task is to show that that is an illusion. I suggest a third reading, that Kant was recognizing (if only briefly) the reality of unconceptualized perception.Footnote ²⁷ That does not count as ‘experience’ in his sense, which implies concept-application and judgment (A106). Unconceptualized perceptions may be ‘as good as nothing for us’ in the sense that we cannot use them in conscious inferences, hence not in science. But they can certainly make a difference to us, as shown in our behavioural reactions to flashes, bumps, gestures, facial expressions and tones of voice. So the aim of the Transcendental Deduction can be reformulated as to demonstrate that the categories must apply to our conceptualized perceptual experience. The subjective side of the Deduction can be construed as a priori reflection on what constitutes perception, whether conceptualized or not – which would explain Kant's claim that his question ‘How is the faculty of thinking itself possible?’ is not, after all, the ‘seeking the cause of a given effect’ (Axvii).

4. Three levels of synthesis

Let us now examine the three syntheses in the A Deduction.Footnote ²⁸ In the first, ‘the synthesis of apprehension’, his main point seemed to be that all representations including perceptions occur successively in time. He talked of ‘the mind distinguishing the time in the succession of impressions on one another’, and declared that ‘in order for unity of intuition to come from this manifold (as, say, in the representation of space), it is necessary first to run through and then to take together this manifoldness (A99)’. That may make it sound like a conscious, deliberate process of mental activity, as when a biologist takes a number of photographs of whales and tries to work out how many different animals there were. But Kant has said that synthesis is seldom if ever conscious. So I suggest that his synthesis of apprehension should be understood as occurring at an unconscious level, in which there is a succession of sensible impressions (not consciously-noticed sensations) which do not immediately fade and ‘leave not a rack behind’, but their effects persist briefly in the sensory systems and form the inputs to perceptual processing. For example, the rapidly increasing size of a retinal image usually leads to awareness (in both animals and humans) that something is approaching, or being approached (though more rarely, the object might be expanding, like a balloon).

The spatial nature of much sensory stimulation and processing is implied by Kant's parenthetical remark about the unity of intuition in the representation of space at A99. This is exemplified in the two-dimensional patterning of light on the retina and in the slight difference between the two retinal images, and the subtly different air vibrations received by our two ears which enable us to locate objects and sounds in public space. There is also the location of stimulations on our three-dimensional bodily surfaces which led to our tactile feelings of tickles, pressures or heat.

Kant's treatment of the second synthesis, ‘of reproduction in the imagination’, at A100-2 is obscure, despite the unusual flurry of examples to illustrate his abstract theorizing. One obvious point is that the association of representations cannot work unless there are rule-governed conjunctions of the objects or events that cause the relevant perceptions.Footnote ²⁹ Kant seems to have had something more conceptual in mind when he talked of ‘my empirical imagination getting the opportunity to think of heavy cinnabar on the occasion of the representation of the colour red’ – but surely not on any sighting of red, e.g. in an apple or a sunset – perhaps he was thinking of the situation of digging up a distinctively red substance. His other example – of the difficulty that would be presented to empirical imagination ‘if a human being were now changed into this animal shape, now into that one’ – implies that our recognition of instances of kinds depends on there being natural kinds (such as creatures and chemicals) that exhibit stable and repeatable clusters of perceptible properties.

However, recognition of kinds can occur at an apparently non-conceptual level: e.g. gorillas recognize which plants are edible for them, and show them to their young, and an African bird makes different alarm calls for different predators.Footnote ³⁰ Does the bird apply a concept of leopard or eagle or snake when it sees one? That depends on what we mean by ‘thinking’ and ‘concept’. It certainly perceives them, and distinguishes between them in its behavioural reaction, but for Kant thought is conceptual and judgmental, and judgments are expressible in language. Primates and birds do not have anything as rich as a human language, although some of them in captivity have been taught some symbol-use, so it seems there is some empirical shading of the supposedly sharp distinction between animal and human thought and language. Without going further into this much-debated area, we can say that there is a mental level of perceptual tracking of individual objects, and recognition of kinds of object, that lies in between mere associative conditioning and linguistically-expressible perceptual judgment. This seems to fit naturally into Kant's second synthesis, since he reserves concepts for the third.

Kant's third synthesis, ‘of recognition in the concept’, leads him away into a discussion of objects of representation, transcendental apperception, and his notion of transcendental object (A103-110), which is beyond the scope of this essay. At the beginning it is difficult to see how the third synthesis differs from the second, for he talks again of reproduction, and of maintaining the content of thought from one moment to the next, and he appeals to the same example of counting. But he goes on to talk of concepts and ‘consciousness of unity of synthesis’, declaring that ‘it is this one consciousness that unifies the manifold that has been successively intuited, and then also reproduced, into one representation’ (A103). This broaches themes that are more fully developed in the B Deduction, to which I will now turn.

I am suggesting that we construe Kant's ‘three subjective sources of cognition: sense, imagination, and apperception’ (A115) as follows:

1. 1. The reception and brief retention of sensory impressions in the sense-organs.

2. 2. The spatiotemporal patterns of these sensory impressions are processed to form unconceptualized perceptual representations that inform the perceiving creature of the shape and size, distance and movement of objects, and of the direction and movement of sources of sound.

3. 3. Rational subjects like ourselves apply concepts and make judgments about what we perceive, and this involves some degree of self-consciousness.

6. Combination

The Transcendental Deduction in B is resolutely top-down rather than bottom-up (as was the later part in A): it starts from human self-conscious conceptualizing mentality, rather than the lower levels of which we may share with the animals. Here Kant uses the term ‘combination’ more than ‘synthesis’ and I want to ask whether there is any relevant distinction. For once in this paper I will put a more extended portion of text under the microscope, the first two sentences in the opening section in B.

Our sense-organs are affected by a stream of physical stimulations in space and time. But we go beyond such mere receptivity, as Kant now explains using a new term ‘combination’:

It may sound as if ‘combination’ and ‘synthesis’ are to be synonyms, but three significant points are packed into this densely-packed sentence:

1. (a) Combination can be conscious or unconscious. But Kant described synthesis as blind and unconscious at A78/B103, so combination must be either a broader notion that includes synthesis, or something quite distinct from synthesis.

2. (b) Combination can be either of ‘the manifold of intuition’ (synthesizing sensible impressions into the perception of an object), or it can be the combination of concepts. But the latter is significantly different, so what did Kant have in mind? It could be the combination of two concepts into a complex concept (e.g. black cat) or into a judgment (This cat is black, or Some cats are black), or within a non-asserted proposition (If that cat is black, it's unlucky, Is it true that black cats are unlucky?). All these involve ‘acts of the spontaneity of the power of representation’, i.e. of the understanding as opposed to mere receptivity or sensibility. But if ‘combination’ covers all these conceptual cases as well as perceptual synthesis, it is a very ambiguous term.

3. (c) Combination can be either of ‘sensible or a non-sensible intuition’. The latter is the pure synthesis of representations of space and time, introduced in the Aesthetic at A24-5/B39 and A31-2/B4 and mentioned at A77/B103 – but that doctrine is beyond the scope of this essay.

Kant's long sentence ends as follows:

But that doesn't sound quite right, for if combination is a mental activity of combining representations it does not follow that combination is itself a representation – the mind need not represent its own activity (infinite regress would surely threaten). It seems Kant should have written ‘acts’ or ‘functions’ rather than ‘representations’ here.

The synthesis of sensible impressions into perceptual representations of objects typically operates below the level of consciousness, but in this opening section of the B Deduction Kant's focus is on understanding, spontaneity, and judgment, and he is here emphatic that ‘all combination … is an action of the understanding’ (see also B134-5). When he says that ‘combination is representation of the synthetic unity of the manifold’ (B130-1), I suggest we construe this not as the synthesis (or syntheses) involved in unconceptualized perception, but the third kind of synthesis, now labelled combination, that is involved in making judgments. As Kant says, judgment is an affirmation of how things are: it is more than ‘the representation of a relation between two concepts’ (B140), it involves objective rather than subjective validity (B141-2). And now consciousness, or more precisely apperception or self-consciousness, comes into the story in a famous way:

That makes it explicit that he saw human judgment as involving self-consciousness, at least potentially. As we have seen, he tends to dismiss unconscious states as ‘nothing for us’, but this does not mean that they do not exist or that they have no effects on us, only that they do not figure in our conscious cognition.Footnote ³¹ In talking of the I think accompanying our representations, Kant surely had in mind judgments rather than intuitions or concepts, for the phrase belongs in contexts of the form ‘I think that p’ for some proposition p (including the case ‘I wonder whether p’). A cat can see another cat, but it cannot say ‘I see a cat’, or even think ‘I think I see a cat’. Of course we do not explicitly attach the I think to every judgment that we make. The point is only that we must be able to: it is the necessity of a possibility. But why the necessity? I take Kant's point to be that for a judgment to ‘be something for me’ it must be available for me to make deductive or inductive inferences, I must hold it open to confirmation or disconfirmation.Footnote ³² So I must be able to hold together any of my judgments as premises for inference, e.g. if I think that p, and I think that q, I can think that p and q. Moreover I can say ‘I used to think not-p, but I now think that p’, and ‘I think that p, but my wife thinks that not-p’. For any single mental act to count as a judgment, it must be a manifestation of that sophisticated mental faculty that Kant calls the power of judgment (Urteilskraft). And so he argues from his table of the logical forms of judgments (A70-80/B95-106) to the necessity for any conceptualizing mind to be able to apply the a priori concepts that he finds implicit in those logical forms. Inference can thus be reckoned (with only a slight stretch beyond the text) as another species of the genus combination, since from one or more judgments as premises, another judgment can emerge as conclusion.

When Kant states that ‘this thoroughgoing identity of the apperception of a manifold given in intuition contains a synthesis of the representations, and is possible only through the consciousness of this synthesis’ (B133, with my emphasis), we may be tempted to apply this to the first and second syntheses involved in perception. His very abstract formulas like this tend to admit of multiple applications at different levels of mentality. But if we construe ‘representations’ here as referring to judgments rather than perceptions, we can make different and relevant sense of it in this context. If the representations here were sensible impressions, the statement would contradict the fact that the synthesis of them is a process of which we are seldom if ever conscious.

This suggests an important qualification to that striking statement which many interpreters have seen as pivotal to Kant's whole argument in the Analytic:

I am not denying that Kant believed this. It reflects his top-down approach to cognition, enquiring into the necessary conditions of our most reflective, self-conscious, judgmental mental activity (what he called ‘experience’). Thus he saw the forms of judgment as the clue to the categories, and the latter as structuring all our perceptions.Footnote ³³ What I am questioning is the general truth of that latter claim. To be sure, a concept can function both in a conceptualized perception (e.g. the recognition of a cat as such) and in judgments involving that same concept in subject or predicate position (e.g. ‘My cat is female’, ‘That is a cat’). But I suggest that the kind of unification or synthesis or combination involved is importantly different in the two cases, for perception involves the unconscious synthesis of sense-impressions into the representation of an object, whereas judgment involves the predication of concepts in a proposition and the belief that that represents how things are.

So I dare to suggest that we can improve on what Kant has bequeathed us by distinguishing different sorts of synthesis or combination.Footnote ³⁴ Above the level of sensory registration (where mere associative learning applies) we should distinguish the following levels of mental processing:Footnote ³⁵

1. (a) the synthesis of a perceptual representation of an object from a spatiotemporal manifold of sensible impressions (e.g. from sounds of scratching and glimpses of fur to the perception of a creature lurking in the woodshed).

A dog can hear such sounds and have such glimpses (and he detects smells we can't), but he has no concepts of rat or woodshed in the Kantian sense of concept. Therefore (a) needs to be divided into:

1. (a₁) syntheses of unconceptualized perceptions of objects from manifolds of sense-impressions (e.g. by dogs or other animals, or when we see or hear an unidentified something buzzing round our head, or turn up our noses at an unpleasant smell).

2. (a₂) syntheses of conceptualized perceptions of objects from manifolds of sense-impressions (e.g. when we realize that there is a furry creature in the woodshed, that it is a wasp that is buzzing around, or that the drain is blocked).

3. (b) the combination of intuitions and concepts into a judgment about a perceived state of affairs (e.g. that this cat is tortoiseshell in colour – which involves whatever synthesis is involved in recognizing cats as such, as in a₂).

4. (c) Propositional inferences, e.g. in which a judgment about a public objector state of affairs is arrived at from a combination of perceptual judgments, e.g.

   (spatial) Here is a finger-bone sticking out of the soil, and here is a toe-bone a few feet away, so probably they belong to the same skeleton.

   (cross-modal) The teacher hears a rude noise, she sees a smirk on Buggins’ face, and decides he is the culprit.

   (temporal) On CCTV the police see a man entering a house, then a similar-looking person coming out later, and they judge it is the same man.

The distinction I am bringing to the fore here is that between (a₁), the unconceptualized perception of objects, and all the other levels, which involve concepts. Kant is surely right to emphasize the interdependence of conceptualized intuitions (a₂) and judgments (b). But I suggest that we need to extend Kantian theory to recognize the distinction between (a₁) the unconceptualized, non-propositional, perceptual representation of objects (found in animals, infants, and to some extent in adult humans), and everything else, from (a₂) to (c). As we have noted, Burge defends (a₁) as a natural kind in psychology. Bill Brewer has offered a philosophical defence of what he calls the ‘Object view’ of perception against the recently fashionable ‘Content View’.Footnote ³⁶ According to the latter, the nature of perceptual experience is given by its propositional content, whereas according to Brewer it consists in a more primitive, non-conceptual ‘acquaintance’ with mind-independent objects. Perhaps Kant would be happy to find recent psychology and philosophy combining to confirm something that he only rather dimly realized.