Imagination and the Conditions of Knowledge
Imagination and the Conditions of Knowledge
Abstract and Keywords
This chapter critically analyzes Kant's theory of schemata and develops it in a more complete way, with reference to the ontogenesis of experience. It argues that the transcendental schemata should be interpreted as retentive and anticipatory procedures that assist the nascent categories to achieve a basic objective orientation in cognition. It is argued further that these procedures centre on the productive imagination's capacity to model possibilities of temporally successive appearance. It is only through the realization of this disposition that we can explain how concepts in general are able to apply. The claim is here justified through phenomenological examples, and is extended to cover empirical schemata. In particular, it is shown how schemata, of their nature, have an element of generality that makes them homogeneous with both concept and sensible intuition.
The preceding chapter has shown how, for Kant, the productive imagination plays a central role in our knowledge of the objective world and the unity of the self. This role is not one which has been greatly emphasized by other thinkers, so it is worth considering its structure and scope in rather more depth. The task is all the more vital for the present work in so far as subsequent chapters will show its centrality to Kant's aesthetic theory.
There is a logical starting‐point for a detailed investigation—namely, the schematism section of The Critique of Pure Reason (which follows on from the Transcendental Deduction). Unfortunately, the schematism is short, notorious for its obscurity, and has not been given much systematic development even by Kant's more sympathetic interpreters.1 (p.36) To the casual observer, indeed, it may appear to be little more than a remnant of European idealism's more Romantic aspects.
In contrast, I will argue that the schematism is of decisive importance to the formation of objective knowledge and, more speculatively, to the unity of the self. It has a philosophical significance in excess of its historical context and the confines of Kant exegesis.
To show this involves strategies of great methodological complexity. In order to clarify the surprising scope of the schematism's significance one must combine both analytic and phenomenological approaches.
In Part 1 of this chapter, I briefly consider Kant's rather unsatisfactory introductory treatment of the schematism. Part 2 addresses Kant's more detailed account of transcendental schemata, and offers some important refinements to his position.
I then proceed, in Part 3, to develop Kant's notion of the productive imagination in more detail, making use of a phenomenological outline of the ontogenesis of experience. Through this, it is argued that the transcendental schemata should be interpreted as retentive and anticipatory procedures in the productive imagination through which the nascent categories achieve a basic orientation in cognition.
It is argued further that these procedures centre on the productive imagination's capacity to model possibilities of temporally successive appearance. It is only through the realization of this disposition that we can explain how concepts in general are able to apply.
In Part 4, my development of Kant's position focuses on the phenomenology of schematization, and is extended to cover empirical sensible concepts. It is shown how realizations of the relevant imaginative dispositions have an intrinsic general significance which is homogeneous with the generality of sensible concepts. Part 5 extends the importance of schematization to some aspects of the unity of the individual self.
I then proceed to a brief conclusion summarizing the arguments in this chapter, and indicating the general importance of understanding and imagination for this study.
As we have seen, Kant's arguments in the Transcendental Deduction are complex and require a great deal of work in order to be made viable. I shall focus now on two clear emphases in his arguments. The first is that knowledge of an objective world and a unified self presupposes both pure concepts of understanding (or ‘categories' for short) and the second is that the imagination plays a key role in this knowledge.
For Kant, imagination is ‘the faculty for representing an object even without its presence in intuition’.2 It has a productive aspect based on the capacity to generate images per se, and a reproductive aspect based on associational connections between images so generated.
The schematism emphasizes the former, and, in particular, its linkage of attention and retention in relation to the sensible manifold (a topic which I will consider in more detail in Part 3).
Kant begins his exposition by claiming that
He also offers a further example with a few more details filled in, as follows:
In all subsumptions of an object under a concept the representation of the former must be homogeneous with the latter; i.e., the concept must contain that which is represented in the object that is to be subsumed under it . . . Thus 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.3
Kant also describes the schema as a ‘procedure’ or ‘rule of unity’ through which a concept finds sensible application.
The concept of a dog signifies a rule according to which my imagination can specify the shape of a four‐footed animal in general, without being restricted to any specific shape that experience offers me, or any possible image that I can exhibit in concreto.4
But, what distinguishes such a rule from that of the concept on whose behalf it is acting? If we say it is the rule sensibly applied as opposed to abstractly conceived then this tells us nothing, because Kant's whole point is that the schema involves an additional factor which enables (p.38) concepts to be applied to phenomena. The concept must ‘contain’ general characteristics of its object which are ‘homogeneous' with the concept.5
It is striking that Kant's examples centre on a capacity to represent shapes which are sufficient to characterize kinds of spatial object, but not sufficient to present recognizable ‘natural’ individual instances of those kinds. Indeed,
It is clear from these first characterizations that, whatever schemata in general might be, the general shapes and forms of empirical spatial particulars are involved in them.
the schema of sensible concepts (such as figures in space) is a product and as it were, a monogram, of pure apriori imagination, through which and in accordance with which the images themselves first become possible.6
However, Kant asserts rather than explains and argues his points. There is one issue in particular which must be clarified before proceeding further. Earlier I noted that the schema must be an ‘additional’ factor which mediates between concept and phenomenal item. Kant's examples seem to indicate that this is some third term, but if the mediation of such a term is involved this would point towards a fallacy of the ‘third man’ type where the mediating term would require another mediating term in order for it to apply, and so on, in an infinite regress.
Given this, to be charitable to Kant, the mediating factor should be regarded not as some singular schema but simply the way in which different concepts engage the productive imagination in different ways. For example, to schematize the empirical concept dog we do not need some schematic dog‐form as a content of consciousness; rather, the schema consists in the way that the exercise of this particular concept directs the imagination in a different way from what is required by other concepts.
(p.39) On these terms schemata are to be understood as different operational procedures determined by the specific demands which particular concepts make upon the productive imagination in order to be formed or applied. They are, in other words, different ways in which concepts engage the productive imagination per se.
This being said, there are still three major questions left unanswered by Kant's introductory characterizations of the schemata. First, in what sense do concepts ‘contain’ features which characterize their objects; second, in what sense are they ‘homogeneous' with such characteristics; and third, how do schemata based on these qualities act as ‘rules' for concept application?
The bulk of Kant's subsequent exposition addresses the transcendental schemata's role in enabling the categories to be applied. However, it does not answer the aforementioned questions in any direct way. Indeed, he introduces a new factor in so far as the burden of emphasis in the transcendental schemata concerns temporality.
In order to work towards answers to the three question, then, we must first consider Kant's treatment of the transcendental schemata in detail—both expounding them, and, where necessary, revising his position.
For Kant, the transcendental schemata enable the application of those ‘categories' which are the basis of our knowledge of an objective world. The categories consist of unity, plurality, totality, reality, limitation, negation, substance, causality, reciprocity, possibility, actuality, and necessity.
Kant treats the quantitative group of categories—unity, plurality, and totality—in terms of a single schema, since they all pertain to ‘time‐series'. The schema is number. It schematizes the relevant categories through ‘the successive addition of one (homogeneous) unit to another’ and, in this way, exemplifies the fact that ‘I generate time itself in the apprehension of the intuition’.7
The qualitative group of categories—reality, limitation, and negation—is again dealt with in terms of a single schema which pertains to the ‘content of time’. It consists of degrees of reality as (p.40) expressed through variations in the magnitude of sensation. In Kant's words,
the schema of reality, as the quantity of something in so far as it fills time, is just this continuous and uniform generation of that quantity in time, as one descends in time from the sensation which has a certain degree to its disappearance or gradually ascends from negation to its magnitude.8
Kant's remaining categories are given individual as well as group characterizations. They can be summarized as follows.
The relational categories comprise substance, causality, and reciprocity, and concern the ‘order of time’ in the connection of perceptions. Substance's schema is that of the permanent which endures through temporal change. As Kant puts it, ‘in it alone can the succession and simultaneity of appearances be determined in regard to time’.9 Causality is schematized through ‘the succession of the manifold in so far as it is subject to a rule’.10 Reciprocity is schematized as the simultaneous rule‐governed interactions of a substance's accidents with the accidents of other substances.
Finally, the group of modal categories. This consists of possibility, actuality, and necessity. They describe the ‘sum total of time’. The schema of possibility involves ‘the agreement of the synthesis of various representations with the conditions of time in general’.11
On these terms, for example, a thing cannot possess opposite properties at the same moment in time; it could only have these at successive stages of its existence. Possibility is schematized, therefore, in terms of the consistency between the specific properties of a thing and its occupying a specific location in time.
The schema of actuality is more straightforward. It consists of ‘existence at a determinate time’.12 Necessity's schema is also relatively clear in nominal terms. It involves the ‘existence of an object at all times'.13
I shall now analyse Kant's treatment of the schemata in more detail. Consider first the time‐series categories—unity, plurality, and totality. Kant's assertion that ‘number’ is their common schema is curious and not at all compelling. The named schemata characterize different ways in which the relation of parts in a whole might be understood. Numbers, in contrast, mark out quantitative scope of an (p.41) accumulation of units. Whilst this scope might be considered in relation to a whole, it is not a necessary factor per se, in so far as one can often—indeed, mainly—characterize things as unities, pluralities, or totalities, without our attention to their parts having to involve any element of enumeration. It is also strange that Kant takes only one schema to be involved in the time‐series categories.
His position here might be developed as follows.
If something occupies space and time then, qua spatio‐temporal, it is composed of a continuum of parts. If one wishes accordingly to comprehend the character of a specific spatio‐temporal unity, plurality, or totality as given in perception, it follows that one must, in principle, be able to apprehend the continuity of its parts as a continuum, if called upon. This requires at least some procedure of successive apprehension.
In the case of a spatial item or state of affairs, this apprehension is relatively open. The continuity of parts in a spatial item or state of affairs can be scanned successively in any direction. In the case of an event, in contrast, the sequence of its ‘parts' qua temporal must follow an exact order of succession. Its phenomenal continuity is strictly linear.
Now, it is clear that we cannot perform such successive apprehensions of phenomenal continuity for every object of perception: our cognitive capacities would be overwhelmed. However, we at least know that the item's phenomenal continuity can be tracked through successive apprehension, as required. Having such a procedure at our disposal is a condition of our being able to negotiate sensible items and states of affairs, qua sensible.
I would suggest then, that unity, plurality, and totality can be schematized in terms of a single procedure, but that it concerns the successive apprehension of continuity of parts, rather than number.
Let us turn now to the cases of reality, limitation, and negation. Kant's treatment of these time‐content categories is especially weak. The concepts of ‘reality’ and ‘negation’ are fundamental to the possibility of experience (in ways which will become clearer as I proceed) but the categorical status of ‘limitation’ is much harder to justify, and, for present purposes, can be set aside.
It should also be noted that Kant's treatment of a group of categories in terms of a single schema is, again, not compelling. Neither is his characterization of this schema in terms of quantitative variation of the magnitude of sensation, specifically. Further factors are involved. I will now try to develop Kant's approach to take account of them.
(p.42) Rather than treating reality and negation as two separate concepts, it is more useful, phenomenologically speaking, to treat them as paired—since each is involved in the full definition of the other, and the transitions between the two are of the utmost importance in our understanding of phenomenal change. Kant is right to identify the transitional factor here with variation of magnitude, but it has three quite distinctive aspects, each of which can be regarded as a schema in its own right. I shall now consider these.
The first is variation of physical magnitude. Anything which occupies time is capable of being physically diminished in quantity until there is nothing left of it. We move from a given reality to its negation.
However, there is also a qualitative version of this. It arises when some item is added to or subtracted from, in such a way as gradually to change it into some other kind of thing (e.g., when a liquid is ‘watered down’ to such a degree that it becomes no more than water).
A second schema involves variation of perceptual magnitude. It has a spatial and a temporal aspect. Broadly speaking, the closer things come towards us in the spatial field, the bigger they will appear to be, and, correspondingly, the more they move away from us, the more their magnitude will appear to diminish, accordingly.
Likewise with time. The nearer an event is to us in time (either past or future), the more influence it will tend to exert on what we do in the present. And, reciprocally, the further away in time it moves, the less it will tend to exert such influence.
The third schema is the closest to Kant's own characterization. It involves variation of the magnitude of sensation. Over and above the perceptual criteria just outlined it is clear that how things impact on our senses—our feeling of them—can have greater or lesser degrees of intensity according to the nature of the impact and the constitution and sensitivity of our sensory receptors.
The schemata which Kant characterizes individually, are rather more straightforward. That of substance—the permanent which endures through temporal change—pertains to invariance of form throughout the different appearances which an item takes. This invariance centres on features which characterize the phenomenal structure of that specific kind of item. (The re‐identifiability of an item as an individual also involves features which are distinctive to it qua individual, but these empirical properties are not relevant to transcendental schematization.)
(p.43) The schema of causality centres on temporal transformations amongst items and states of affairs. Changes of this kind happen in a rule‐governed way which through the productive imagination's power of attention and recall can be tracked and negotiated before being explicitly recognized through concepts.
In the case of reciprocity, imaginative attention and recall (with the emphasis on the former) are directed not so much towards the temporal trail of events and the anticipation of outcomes but rather towards a sense of how things which are physically proximal can mutually modify one another—in causal terms or through simply limiting each other's space of action.
Finally, the modal schemata. These are orientated towards temporal coherence in the perceptual field. If something is possible, it must have attributes which are consistent with how a thing of that kind is located in time.
A living being, for example, cannot come into existence at its death, and then move through time towards its birth. Likewise, a pebble cannot be physically transformed into particles and then, a moment later, be reconstituted as a pebble without any intervening process of physical action upon it. The possibility of things is dependent on their properties being acquired consistently with the passage of time.
The schema of actuality is simpler than this. For, if something exists, it must occupy some portion of space at some time or other. (A world of sounds cannot be used as a counter‐example here, since sound is intelligible only in terms of the perturbation of space‐occupying phenomena.)
Necessity's schematization is a little more complex. It is difficult to make sense of Kant's notion of necessity as ‘the existence of an object at all times'. However, such necessity can be comprehended in terms of factors which constrain the existence of all spatio‐temporal items and states of affairs, qua spatio‐temporal. For example, no material body can occupy the exact same spatial coordinates as another at the same time. No material body can move from one part of space to another without continuous traversal of the intervening regions.
Having expounded and developed Kant's account of the transcendental schemata, I shall now consider the productive imagination in more detail, with a view to presenting his general theory in a viable form. This will involve a phenomenology of the ontogenesis of experience which brings into play a key factor which Kant neglects—namely, anticipation.
As we have seen, for Kant, the imagination is a capacity for representing things which are not themselves given in immediate perception. This capacity is of the most vital importance. For, whilst the imagination is dependent on what is given in time and space, in Kant's terms, time and space themselves are intimately connected with the workings of the productive imagination. Indeed, as we have seen, he even implies that imagination is somehow involved in the actual production of space and time.
The point of philosophical substance here does not require that we go as far as Kant on this last point. There is a more economical alternative which holds that imagination is simply a necessary condition for knowledge of spatio‐temporal phenomena. This is because the embodied subject exists in space, and perceives things in space, through movement (be this from one position in space to another, or a simple process of eye‐movement in scanning an object).
By definition, movement of any sort entails temporal succession. To experience such succession in a coherent order requires that our sense of what is immediately present must be informed by a sense of what has gone before. We must be able to retain some of the previous relevant perceptions. Attention entails retention.
However, there is another factor not emphasized by Kant, but which must also be brought in—namely, anticipation. To see why this is so, I shall first undertake an outline phenomenology of the ontogenesis of experience.
All animals can deal with their environments in the sense of being able to negotiate specific kinds of stimuli, in specific ways, for specific ends. Such behaviour is not based on developed conceptual abilities and rational planning (which are enabled through mastery of language) but neither is it mere instinct in the way that less‐developed life‐forms (such as insects) negotiate their environments.
One might say that even though animals do not possess objective knowledge they do have, nevertheless, objective orientation. They can negotiate their environments in a way that is informed by some sense of (p.45) what the world's objective physical features allow or disallow vis‐à‐vis the creature's own capacities.
Human beings, of course, have objective knowledge of their world. However, knowledge and rational planning do not simply happen in an instant. They have to be learned. And, if the linguistic abilities on which such knowledge is founded are to develop, it is presupposed that some stable proto‐cognitive factors enabling such development are already in place. There must be some form of objective orientation.
In the previous chapter, I linked this to the categories in their nascent pre‐reflective form. The categories are rules for recognizing objective structure in sensible manifolds. But, if such rules are to get a purchase on phenomena, they must, at the same time, involve ways of negotiating the manifold in terms of coherent patterns of temporal succession. This is the function of the transcendental schemata.
Such procedures do not exist independently of the categories. One presumes, rather, that the urge to cognitive unity embodied in the different categories is what orientates us towards specific patterns of temporal succession which have to be dealt with if the nascent categories are to develop and issue in full objective knowledge. The transcendental schemata, in other words, are specific ways in which the categories determine productive imagination.
Again, a wealth of striking relevant empirical evidence has been summarized by Gallagher. He notes that neonates have
sensitivity to motion parallax—when the infant moves its head, nearer parts of a three‐dimensional object seem to move more than further parts. [Indeed,] Neonates are capable of discriminating between different geometrical shapes, such as triangles, squares, and circles . . . Their vision is also characterized by shape constancy, that is, across changes in orientation or slant neonates are capable of recognizing the real shape of an object . . . Neonate vision also shows feature constancy, that is, the ability to recognize invariant features of an object across certain varying features, such as moving versus stable objects.14
The recognitions of different patterns of invariancy, variancy, and unity across time suggest that the basis of categorial discriminations and correlated transcendental schemata as perceptual tracking procedures are in play—however crudely—more or less from birth. The neonate is, as it were, prewired to negotiate experience's character as both temporal and centred on phenomena which are spatial, or inseparably related (p.46) to space‐occupying bodies (as in the case of subjective states, such as thought or feeling).
Now, whilst, for Kant, space and time play a receptive role in our cognitive framework, they are not wholly passive. Their properties determine how the items which are presented in them will appear. Of course, all perceptual items and states of affairs are different at the individual level, but there are basic features vis‐à‐vis space occupancy and its temporal conditions which determine the structure of their appearance in general terms.
In relation to the ontogenesis of experience, the mastery of these structuring features involves learning operative skills in relation to perceived items and states of affairs. These build on the innate propensities summarized by Gallagher, and involve a capacity to retain some sense of the previous appearances of the item which is currently being attended to.
However, the role of temporality in the gradual achievement of objective knowledge is surely as much, if not more, future orientated. How we negotiate the manifold is informed by anticipations of how, under different conditions of temporal succession (determined by the perceiver's orientation and movement, or the movement of objects), certain aspects of appearance will develop in such and such a way but not others, or will issue in such and such outcomes but—again—not others.
Through attention informed by retention, and anticipation, the child learns to negotiate the ‘thingliness' and eventfulness of the world in a basic way even though it has not yet learned to understand these in terms of explicit categories and empirical concepts.
Basic sensori‐motor negotiations of this kind centre on such things as constancy and invariance of spatial form and size (and, thence, identity) in things across variations of light conditions, and across different distances. We must comprehend the regularity of events and become aware of the capacity of some things and events to act upon us, and we upon them. We must form a sense of which things can be dealt with as discrete unities, and which things are easier to negotiate as aspects of some greater whole.
The basic levels of recognition involved here are the nascent categories. But, in order that conditions for such basic idioms of recognition are in place, we also need to track the relevant salient patterns of temporal succession in the sensible manifold.
As we have seen, Kant's transcendental schemata centre on the successive perception of extent or duration, substantiality, causal (p.47) regularity, variations of magnitude, and patterns of temporal consistency amongst phenomena. These are precisely the temporal ‘tracking’ procedures—involving retention and anticipation—which the achievement of objective orientation requires.
They have their origins in perceptual habits acquired through the gradual coordination of the body's mental and sensori‐motor powers in a unified field. For example, through repeated handling of such things as toys, and interactions with other people, the child becomes able to anticipate what their hidden perceptual aspects might be like, and finds ways to disclose them or have them disclose themselves, if not immediately accessible.
Likewise, it can retain a sense of its own past, and anticipate its possible future movements on the basis of these interactions. It learns an operational schema of possibilities in terms of how specific things, and their contexts of occurrence and location, have and might be expected to appear—but without yet identifying these as appearances of individual things or particular contexts of occurrence.
If this approach is correct, then we might characterize the transcendental schemata more specifically, as pre‐reflective procedures which involve a sense of how phenomena have appeared, and anticipate how they might come to appear under specific conditions of temporal succession.
The nascent categories provide the recognitional dimension of objective orientation, whilst the transcendental schemata are the temporal tracking procedures—specific demands exerted on the productive imagination—which enable such recognition. Without these procedures we could not achieve objective knowledge because we would have no perceptual sense of there being any objects to know.
On these terms, then, the fact that a child has learned the ‘thingly’ character of things, and the eventful character of events, in these operational terms means that its ability to form explicit category and empirical concepts is in place. The acquisition of explicit concepts through language both feeds into and extends this basic objective orientation.
Now it might be admitted that retentive and anticipatory procedures of the kinds described are, indeed, necessary for the formation of objective knowledge. But it might also be argued that this position does not, of itself, justify Kant's connection of these with imagination specifically. Might they not, for example, be explained sufficiently as a function of attention informed by memory traces alone? Why should imagination be invoked at all?
(p.48) An answer has already been given. In humans, objective orientation develops into concept acquisition. Since concepts, by definition, must apply to individuals in places and times beyond the limits of the present perceptual field, this means that, in order to form and apply them, human anticipatory procedures must be informed by a broader representational power—one which enables the projection of situations beyond the contents of the present perceptual field. Memory traces are simply not enough here.
And neither can the requisite trans‐ostensive power be explained in terms of concepts of a more rudimentary ur‐kind. At some point, an extra‐conceptual factor will have to be invoked so as to explain the trans‐ostensive leap which enables concept formation. If this were not the case, we would be caught in a vicious regress of concepts required to ground conceptuality and concepts to ground the concepts which ground conceptuality, ad infinitum.15
There is, as far as I can see, only one way to avoid such regress. It is to link human anticipatory procedures to imagination's capacity to produce quasi‐sensory images. There is no other candidate for this role available.
Given some object, one can recall its previous states, and anticipate that it has hidden aspects or possible relations which will become perceptually available in correlation with the percipient's or object's changes of position. One does this in the fullest sense through the occurrent projection in imagination of their possible regularities, persistencies, and variations of appearance.
However, two important qualifications should be made to this. First, such imaginative activity does not reproduce the hidden or successive aspects or past appearances of its intended object in exact detail. It is literally a quasi‐sensory schematization of possibilities of appearance. It is, in important respects, selective.
If, in contrast, the relation between imagination and its objects were one of exact correspondence, then images would obtrude upon the immediately given in a way that would collapse the distinction between actual and possible immediate perception, and thus render our perception of reality incoherent.
Second, as the subject becomes habituated to objective past or future vectors of phenomenal change, his or her perceptual negotiations (p.49) of them rarely require the occurrent exercise of imagination. Such occurrent employment settles—through sustained perceptual familiarity with the relevant dimensions of appearance—into a disposition. This means that it is not made occurrent in every cognitive act, but can be so realized as perceptual circumstances demand.
I am arguing, then, that in order to avoid infinite regress in the explanation of concept formation's trans‐ostensive aspect, we must suppose the mediation of imagination as a disposition to represent that which is not given in perception. This disposition (directed towards specific aspects of past or possible future appearances) is the basis of the transcendental schemata.
We now have a general answer to the first of the key questions noted in Part 1. It is that of the sense in which schemata are ‘contained’ within concepts. In the case of transcendental schemata, we see that they are not ‘contained’ in the sense of being occurrent in every concept‐application. Rather, they are dispositions which allow the tracking of specific general vectors of temporal succession in phenomena, thus rendering the unity of the manifold amenable to recognition. The very fact that they project phenomenal characteristics beyond the immediately given perceptual field means that it is through schemata that we become actively aware of the world's rich and enduring phenomenal structures and processes. They orientate us towards that fabric of appearance which (as we shall see in more detail in the next chapter) is the focus of the pure aesthetic judgement.
Further analysis of schemata‐as‐dispositions can also answer the other two outstanding questions from Part 1. These concern the sense in which schemata are ‘homogeneous' with sensible concepts and the sense in which they provide ‘rules' for such concepts.
To formulate these answers, however, we must offer a more detailed phenomenology of imagination's occurrent character in the schematization of concepts. This will also allow the approach to be extended to the schemata of empirical sensible concepts.
To have a sensible concept in the fullest sense entails comprehension of its extensional scope—that is, the fact that it can be applied not only to directly perceived items, but also to different individuals which share (p.50) the same properties, and which are distributed across, and perceivable by, different observers, at different times and places.
Objective orientation achieved through the nascent pre‐reflective categories and their transcendental determinations of imagination already goes some way towards explaining how extensional scope can be understood as a basis for explicit individual concepts.
For if, say, a child learns the ‘thingliness' and eventfulness of phenomena by habitual procedures, it is reasonable to suppose that, as a factor in such activity, it will, at the same time, learn some individual characteristics of specific kinds of things and events, and, in this way, be prepared for their explicit, conceptually mediated recognition.
These can involve instances of the categories, or of empirical sensible concepts. In either case, we are dealing with schematization that is more specifically focused than in the transcendental schemata's pre‐reflective orientational role.
For example, when a ball is encountered, the child anticipates, on the basis of habitualized experience, that it will easily move if pushed, and will rebound from things if pushed hard enough.
But, in such experiences the child does not yet distinguish between individual balls and ‘ballhood’ per se. There is a general factor of a specific kind embodied in the individual encounters, but this has not yet been made the basis of an explicit concept.
The potential of such a tacit generality to be raised to explicit concept status is augmented by another aspect of the transcendental schemata. Earlier it was argued that without such procedures there would be no way of explaining how the trans‐ostensive dimension of concepts could emerge. Such a dimension is basic to the comprehension of any concept's extensional scope.
Now, as retentive and anticipatory procedures, the transcendental schemata are simply habitual ways of dealing with structural features of things and events. However, the imaginative dispositions in question must, qua dispositional, be capable of sometimes being realized occurrently. The phenomenology of this is extremely instructive.
For example, suppose that, given a partial perspective on a building, one imagines what its hidden aspects and interior might be like. And, suppose also that one has never seen this actual building before. Under such circumstances, one's imaginings of the aspects and interior are consistent with the given stimuli, but whether or not they accurately represent the hidden aspects is another matter entirely.
(p.51) However, the decisive point is the mere fact of some general phenomenal consistency in such imaging. They are schematic procedures which are consistent with enough phenomenal features to define the kind of thing in question. This is all that is required in order to negotiate the ‘thingliness' of the thing in basic terms.
Indeed, in its productive form, imagination just is a quasi‐sensory capacity for the modelling of possible appearances of specific kinds of sensible item. It need not be constrained by exact correspondence with existing sensible items, and, indeed, cannot be—given the schematic and stylized character of images per se (a topic I will return to in more detail a little further on).
Through this imaginative modelling, our sense of the phenomenal world's perceptual continuity within and beyond the immediate field will never have gaps. Given some portion of reality, we can imagine how it might be configured in areas to which we do not have immediate perceptual access.
Indeed, as long as an image sequence constitutes a logical schematic continuation of the given and presents how it might appear or might have appeared, then, in a key sense it does not matter if our subsequent explorations or recollections show the imagery to be not entirely accurate. In such cases, even though our modelling may be factually mistaken, it is at least coherent in having articulated a possibility of appearance.
This primacy of general consistency over exact correspondence in the exercise of the transcendental schemata opens up a broader possibility. For, whilst such modelling of appearances is learned initially in an ostensive context, its general character means that it can gradually be used autonomously—that is, as a basis for modelling kinds of phenomenal item or event other than those given in the immediate perceptual field.
For the child, encounters with perceived items may lead to daydreams or fantasies involving the imaginative fabrication of things or situations other than those which are present to perception. Through this, the child models possibilities of appearance which are not derived from direct empirical experience but which, nevertheless share some general phenomenal traits with it.
Even if its imaginings are of specific individuals, these contain general characteristics of shape, size, and detail which can also be developed easily into the imagining of other individuals of the same kind—even though these ‘general’ individuals are not clearly separated from the one originally represented.
(p.52) Interestingly, there is a bias towards generality even in the case of specific imagined individuals. Suppose, for example, that a child imagines its mother. One presumes that if the child's linguistic capacities are developed the image will be able to satisfy specific descriptions—e.g., ‘mother when she meets meet me from school’.
But, if the image is just of mother per se (and especially if the child's linguistic capacities are not developed) it will involve some of her identifying general phenomenal characteristics, but not as given under some exact time and place of perceptual encounter.
Indeed, even in the case where the child can imagine ‘Mother when she meets me from school’, various kinds of core phenomenal content will be consistent with the image's realization. If the image is not a sustained sequence of imagery, then it must have appropriate synoptic/schematic features. Its selectivity will demand an element of generality.
One does not mechanically reproduce past appearances, one generates them as consistently as possible with the specific description of them. If the description entails more than one event, then the generalizing factor will be all the more pronounced. But, even in recollections of a single event, interpretation rather than exact mechanical reproduction is involved. (I will return to this point in more detail in Part 5.)
In this way, then, possibilities of appearance come to be modelled with a significant degree of autonomy from the actual. Such autonomy means that the child's cognitive power is now extended radically, and that its introductions to language are informed by a decisive trans‐ostensive representational factor.
Reciprocally, the acquisition of language allows key recognitional distinctions to be made at the level of imagination and perception. Specifically, the child is able, gradually, to distinguish between imaginative fantasies per se and imagined real possibilities and between kinds of thing, their individual instances, and the relation between these elements.
Imagination's capacity for modelling appearance in general is, then, complementary to explicit concept formation. This encompasses not only the general sense of ‘thingliness' and eventfulness that is directly involved in category formation, but extends also to our acquisition of empirical sensible concepts.
A further key aspect of productive imagination is also relevant to both these dimensions of conceptualization. For the extensional scope inherent to the notion of a concept means not only that a set of (p.53) phenomenal properties can be instantiated by many individuals in different places and times, it means also that these are, in principle, accessible to different observers, under different perceptual conditions. If a concept is to be comprehended qua concept, these possibilities of different observational conditions must inform its articulation.
Now, in schematizing the previous or hidden aspects of a given object, there is, of necessity, qua schematic character, an element of selective interpretation, and this is true even if the object's hidden aspects are ones which have actually been experienced previously.
The element of schematic selectivity is, of course, even more pronounced in those cases of imagining items or states of affairs which are not immediately given, or of which we have not had direct previous experience.
This leads to the decisive point. The very fact that phenomenal items and states of affairs are schematized through selective interpretation has great significance vis‐à‐vis objective cognition. It means that the schema characterizes its object from the imaginer's point of view. There is a core of objective properties which determines what is selected, but how these properties are characterized is down to the individual imaginer's subjective perspective upon them.
Given, therefore, that in imagination, conceptual content cannot be separated from subjective orientation, the image has the potential for exemplifying the general fact that such a content is encounterable under different perceptual conditions.
This point should now be linked to the other key intrinsic features of the productive imagination's occurrent imagery—namely, its tacit concrete universality, its orientation towards general phenomenal consistency rather than exact correspondence with the object, and its potential to be used independently of that which is given in immediate perception.
If we connect all these points, it is clear that the occurrent exercise of imagination, as such, has an inherent dimension of generality. This answers the second question left unanswered in Part 1. Sensible concepts are ‘homogenous' with general features of their objects because they are enabled through the mediation of imagination—a cognitive capacity whose occurrent exercise, itself, tends (through its schematic character) towards some generalization of the intended object's phenomenal features.
It is surely this homogeneity which provides a basis for learning of general linguistic rules and which, reciprocally, the acquisition of language refines (thus enabling the full formation and application (p.54) of concepts based on a sense of their extensional scope). Imagination provides the mediating trans‐ostensive dimension which allows concepts to be articulated.
This relates directly to the third question left unanswered from Part 1—namely, the sense in which schemata are ‘rules' for applying concepts. If my arguments are correct, it is clear that schemata are necessary to the formation of concepts. However, it is less clear as to how they might be regarded as rules for the application of concepts, in the mature language‐user.
It comes down to this. Could we recognize, say, a saucer or a dog as an instance of the relevant concept, in the absence of a disposition to schematize it in terms of sensible core properties which define the concept in question?
The answer is ‘no’. For whilst schemata need not always be involved occurrently in the conceptualizing of sensible items, every such concept must be, nevertheless, in principle schematizable—that is, it is entailed that if one has a concept of a sensible item one must be able also to imagine what an instance of that concept might be like in general terms.
This is true even of concepts which one has never actually seen an instance of, or of fictional entities. One may know the notion of a ‘hippogryph’ only as a mythological term per se. But if—through description—we know what kind of thing such an entity is, then it becomes possible to schematize how it might appear even though we have never seen such a creature in actuality.
The point is, then, that for a language user to comprehend a sensible concept as such, the concept must be schematizable—the user must, through imagination, be able to model how an instance of it might appear in general sensible terms (irrespective of whether or not such an instance is present in the immediate perceptual field).
If, in contrast, a subject could only negotiate the basis of such a concept's appearance through recognition of immediately given instances of it in concreto, then that subject would, surely, not yet have mastered the relevant concept in its full extensional scope. Its recognitional capacities would be more animal than human.
To understand a sensible concept, in other words, cannot be separated from the capacity to schematize an instance of what it might be like qua sensory—however bare such a schematization might be. One must know it as a possible appearance—that is, in terms of the sensible conditions under which it occupies space and time.
(p.55) To reiterate, then, the application of concepts to sensible manifolds does not, of itself, necessitate the occurrent mediation of schemata, but since schematization is so intimately bound up with the trans‐ostensive basis of concept acquisition and application, one cannot separate the having of a sensible concept from the dispositional capacity to schematize it in imagination (if cognitive circumstances demand).
Schematizability rather than schemata, as such, provide a rule—or, better, a criterion—for comprehending the general sensible scope of a concept's application. Even though the criterion here is psychological, it could be also be expressed in public terms if the subject were able to follow commands of the form ‘draw a picture of an x—not a particular x, just an x’. Through being able to do this, the subject would provide publically accessible evidence of knowing ‘what it is like’ for the term to apply in general sensible terms.
It should be emphasized that as well as serving as a logical criterion of the subject's comprehension of a concept's sensible scope, the schemata also has a much more mundane, but phenomenologically vital role.
Our everyday cognitive life is replete with recognitional acts involving concepts. And, whilst perceptual contexts rarely demand that the correlated dispositions to schematize are exercised occurrently, they often are—purely by chance or association. We may, for example, be thinking of some kind of thing and find that an image exemplifying some characteristic appearances of things of that kind happens to come to mind.
Happenstances of this kind are quite common and important. For, through them, something which is a necessary formal condition of cognition comes also to enrich the general phenomenal texture of subjective experience. The significance of schematization vis‐à‐vis self‐knowledge can, indeed, be taken much further. It is to this possibility that I now turn.
As we have seen, to schematize is to be selective, but with a subjective emphasis. Even in their exemplification of general features schemata are manifestly stylized—i.e., have a particular character.
This has some remarkable implications for self‐knowledge. Kant does not consider these, but they are generally consistent with his approach. (p.56) They converge on what follows from the intrinsic connection between stylization, imagery, and memory.
It is all too easy to think of memories as faded copies of ‘original’ experiences—i.e., as their decayed indexical residue. However, this is a woefully incorrect model. To see why, we must first note a distinction between ‘fact memory’ and ‘corporeal memory’. The former is a recall based exclusively on description whilst the latter involves memories with some sensible—that is to say, imaginative—content.
As our experience increases, a specific factual description may be rendered less amenable to present recollection, but success in recollecting it does not require that we sort through our past experiences until it is found. What happens, rather, is that what is accessible to memory can be schematized in a way which is consistent with what that memory describes in factual terms. Such corporeal memory is projected on the basis of our present interests. Its style is indelibly shaped by these.
It is this very factor which is central to personal identity. Memory is at its most real for us when it is projected as image rather than conceived as fact alone. However, because of the gap between the image and its referent, this means that the present context of projection must interpret the memory from its own point of view. How the past appears is thereby inseparable from the present.
And, of course, there are many areas of our past of which we may have some factual knowledge without being able to recall in sensible terms what the experiences in question were like. However, in such cases, we can project imagery which is consistent with what we know, and in this way even unknown aspects of the past are filled in through the fact of their being schematized in relation to present knowledge.
This might serve as the basis of a more general theory of personal identity. Indeed, some unexpected further elements can also be filled in by reference to the objective dimension of experience.
As an entrée to this, however, we must draw another inference from the nature of the image per se, which Kant does not. It concerns an implicit reference to the personal identity of anyone who schematizes.
If, as we have seen, any image is stylized, then it exemplifies (to some degree) a personal perspective on things. In the previous few arguments, I have emphasized this perspective as being interpretative and present‐orientated. But, there is another implicit aspect which centres on perceptual possibility.
To schematize is to exemplify conditions which are consistent with a concept's sensible application. If this is so, then, reference to a possible (p.57) observer is entailed in so far as sensibilia are, by definition, in principle accessible to perception.
It is important to stress the ‘possible’ aspect here. For the schema‐product does not have to exemplify actual concept instances which we have actually experienced. The general observer implied in schematization, is, rather, one which, by definition, notionally includes the one who performs it. Even though he or she does not intend such specific reference in order for the schematization to occur, it is nevertheless phenomenologically entailed.
There is, therefore, an inescapable ‘possible for me’ clause built into schematization. It is not just that it is this particular ‘I’ who projects the schema, but also that what is so projected is one aspect of a world of possible concept occurrence and application which must include myself qua concept‐user.16 Indeed, the stylized character of the schema will tend to underwrite the implicit subjective reference factor here.
This intimacy of subject and concept achieved through schemata has a further decisive significance.
Earlier, I emphasized the necessary cognitive role of the transcendental schemata—as an enabling factor in the formation of objective knowledge. Given, therefore, the foregoing account of the nature of imagery, this means that in so far as objective knowledge necessarily involves schemata, our conception of the object is inseparable from the having of possible personal perspectives upon it, even though it does not explicitly draw upon our actual experiences of it. A disposition to model the scope of concepts will, when made occurrent, issue in exemplars not only of the concept in question but of a personal way of modelling it.
Now, for Kant, the kind of connective ‘synthetic’ activity bound up with concepts of self and world per se has a basically formal significance. It involves a unified self as a necessary condition of objective knowledge through its involvement in the application of the categories (the doctrine of ‘pure apperception’, as Kant calls it).
However, the foregoing analysis has opened out an entirely new prospect. For it is clear that the imagination's schematizing activity exemplifies the spontaneity of the self qua individual cognitive agent. By understanding transcendental and empirical knowledge of objects and (p.58) states of self‐knowledge as necessarily linked to this activity, we link them to the exemplification of the individual self as well as to subjectivity in formal terms.
The ramifications of this are of great interest, both within Kant's philosophy and in broader terms. In relation to the former, the schematism may now offer the basis for a rethinking of Kant's notoriously opaque conception of personal identity.
At the very least, it certainly extends and consolidates his basic strategy in the transcendental deduction—i.e., showing that subject and object of experience are reciprocally dependent through their involvement in the application of the categories. His emphasis on schemata fills in key details, and my development of it gives further credibility to its overall claims.
In more general terms, there is something about the role of schemata which links the past, present, future, and counterfactual states of a subject. Schematizing is done in the present and, whatever its content, exemplifies something of the personal perspective of the cognitive agent.
Kant notes that it is not the business of imagination to furnish intuitions but to organize the manifold of sensible perceptions.17 But, in the schemata we have a factor necessary for the objective organization of experience which acts upon time through, in part, the subject's unique perspective upon it. The exemplification of this might not amount to a perception of the self in Kant's or any other terms, but it may be about as close as we can get. There is perhaps room for an exemplificational theory of personal identity to be constructed around it.
I shall not pursue this possibility here. The key point is that if this analysis is correct then imagination is not only a necessary condition for knowledge of an objective world but, through its role in this, is also implicated in the unity of the self.
In this chapter, then, I have explored the role of the schematism in that reciprocity of objective and self‐knowledge which is achieved through the categories. In particular, we have seen how the transcendental schemata consist of tracking procedures based on dispositions to imagine temporal (p.59) successions of appearance. They are procedures arising from demands made by specific concepts on the productive imagination.
Through such demands, the unity of the manifold becomes accessible to recognition. Without these dispositions and their occurrent realization as quasi‐sensory images, there would be no viable explanation of how concepts acquire their trans‐ostensive scope.
In the course of discussing the schemata of empirical concepts, I showed further that schematization through quasi‐sensory imagery has a number of intrinsically general characteristics. In concert, these are highly complementary to concept formation. It is, indeed, this intrinsic generality which gives the schema's occurrent realizations some homogeneity with that of sensible concepts.
Finally, I argued that schematizing activity might also be taken to encompass key aspects of memory, and to exemplify the activity of the self at the level of individual cognitive being as well as in terms of its formal unity.
There is a supremely important conclusion which must be reiterated for the purposes of subsequent chapters. It is that understanding and imagination (and the relations between them) are central to objective knowledge and its correlation with both the unity of the self and the character of individual subjectivity.
It should also be recognized that the relation of understanding and imagination means that cognition is not simply a process of classification through judgement. In order for any empirical judgements to take place, we need the categories and those perceptual tracking procedures which are provided by the schemata. As we have seen, schematization is a selective and stylized process. Its projection of content in accordance with a concept is an inherently creative one. Overt acts of recognition involve a correlation of phenomenal unity and diversity, achieved through understanding and imagination.
Given this, it is hardly surprising that the cornerstone of Kant's aesthetic theory turns out to be the heightened and creative interaction of understanding and imagination. His aesthetics reaches deep into the factors which enable knowledge of self and world. In particular, the role of the categories and transcendental schemata is of decisive significance in mediating the aesthetic judgement, as will be shown in many examples in the following chapter. It is to this rich cognitivist aesthetics I now turn.
(1) Strawson's influential paper on ‘Imagination and Perception’ re‐established the intellectual respectability of imagination's role in cognition, but did not give centrality to the role of schemata. For an accessible version of the paper, see Kant on Pure Reason, ed. Ralph Walker (Oxford: Oxford University Press, 1982), 82–99. A more sympathetic treatment of the schematism itself can be found in chap. 8 of Henry E. Allison, Kant's Transcendental Idealism: An Interpretation and Defense (New Haven, Conn. and London: Yale University Press, 2004). Another sustained treatment of the subject is in chap. 2 of Sarah Gibbons, Kant's Theory of Imagination: Bridging Gaps in Judgement and Experience (Oxford: Clarendon Press, 1994). Another probing account is Michael Pendlebury's ‘Making Sense of Kant's Schematism’, Philosophy and Phenomenological Research, LV, 4 (1995), 777–97. Pendlebury has also offered a very useful treatment of related issues in a paper on ‘The Role of Imagination in Perception’, South African Journal of Philosophy, 15 (1996), 133–7. My differences with Allison, Gibbons, and Pendlebury centre, primarily, on the fact that they do not adequately address the phenomenology of the schema per se—that is, the relation between the schema's status as a rule and the imaginative product which that rule generates. I have similar problems with Béatrice Longuenesse's otherwise brilliant, sustained analysis of the logic of the schematism (and related issues) within the Critical corpus, which is distributed throughout Part 3 of her Kant and the Capacity to Judge (Princeton, NJ: Princeton University Press, 1998). For a survey of the general relation between imagination and knowledge (which picks up themes from the Transcendental Deduction), see chap. 2 of Eva Schaper's Studies in Kant's Aesthetics (Edinburgh: Edinburgh University Press, 1979). Schaper does not, however, offer much elucidation of the schematism either here or elsewhere in her book. Jane Kneller's recent study Kant and the Power of Imagination (Cambridge: Cambridge University Press, 2007) also has surprisingly little to say about the schematism, preferring instead to view Kant in a broader historical and intellectual context.
(2) Kant, The Critique of Pure Reason, trans. Paul Guyer and Allen Wood (Cambridge: Cambridge University Press, 1998), A137/B176, 256.
(5) Allison, for example, tries to help Kant by distinguishing between ‘discursive’ and ‘perceptual’ rules, linking schemata to the latter. On these terms, the schema ‘functions to process the sensible data in a determinate way, thereby giving one a sense of what to expect on the basis of certain perceptual “clues”. For example, on seeing the front of a house, one naturally expects that it will have sides and a back with appropriate “house‐ish” features. Rules of this sort are intimately connected with the perspectival nature of perception and, therefore, with the imagination’ (Allison, Kant's Transcendental Idealism, 210). Allison is right to link schemata to perspectival factors. However, if the ‘sense of what to expect’ here is to be more than a mere abstract expectation (i.e., the mere idea of a hidden aspect), then it must have some sensible content, that is have an occurrent image character or a dispositional orientation grounded on such occurrent imagery. It is this factor on which the burden of my own analysis falls.
(6) Kant, Critique of Pure Reason, B181/A142, 273–4.
(8) Kant, Critique of Pure Reason, B182/A143, 275.
(14) Shaun Gallagher, How the Body Shapes the Mind (Oxford: Clarendon Press, 2006), 164–5.
(15) For a lucid summary of the problem here, see Gibbons, Kant's Theory of Imagination, 8.
(16) This point is also recognized (but without being significantly developed) by Wilfrid Sellars in his ‘The Role of Imagination in Kant's Theory of Experience’, in Henry W. Johnstone (ed.), Categories: A Colloquium (University Park, Pa.: Pennsylvania State University Press, 1978). See esp. pp. 236–7.
(17) See Kant, The Critique of Pure Reason, B179/A140.