Wilfrid Sellars

Editor's Note: This article was originally published in Categories: A Colloquium, edited by Henry W. Johnstone, Jr. (Pennsylvania State University, 1978), and appears here with the kind permission of Professor Johnstone. Edited in hypertext by Andrew Chrucky.

1. My aim in this paper is to give a sympathetic account of Kant's theory of the role played by what he calls the productive imagination in perceptual experience. My method, however, will not be that of textual exegesis and commentary, but rather that of constructing an ostensibly independent theory which will turn out, it just so happens, to contain the gist of the Kantian scheme. Proceeding in this way will enable me to avoid the tasks involved in coping with Kant's terminology, architectonic, and polemical orientation. By concentrating attention on the subject matter itself, this approach will make possible a relatively brief treatment of what would otherwise be a time-consuming enterprise.

2. By referring to the theory I am about to construct as "ostensibly independent," I also mean to imply that although I shall stick reasonably close to what I think to be the truth, I shall not be above warping and slanting the argument to fit the role of a sympathetic interpretation of the Critique. The extent to which I succeed in capturing the spirit of Kant's thought must be measured by the degree to which it illuminates the letter of the text.

3. Our access to the external world and to the nature and variety of the objects (in a suitably broad sense) of which it consists is through perception. Phenomenological reflection on the structure of perceptual experience, therefore, should reveal the categories, the most generic kinds or classes, to which these objects belong, as well as the manner in which objects perceived and perceiving subjects come together in the perceptual act.

4. I shall therefore begin my reflections on Kantian themes with a careful account of the relevant features of perceptual (in point of fact, visual) experience. An initial survey will provide a framework of working distinctions which will subsequently be refined. These distinctions, in one form or another, are familiar tools of the philosopher's trade. It is the subsequent refinements that will lead into the arena of controversy.

5. In the first place there is the distinction between the act of seeing and the object seen. Visual experience presents itself as a direct awareness of a complex physical structure. It also presents itself as having a point of view, as perspectival. Opaque objects present themselves as endowed with facing colored surfaces. I do not mean by this that they present themselves as complex structures of color expanses (visual "sense data"), but rather that they present themselves as three-dimensional physical objects which stand in such and such relations to each other and to the perceiver's body.

6. In the second place, there is the distinction, already alluded to, between the objects perceived and what they are perceived as. Thus in veridical perception occurring in optimum circumstances--I shall have nothing to say about illusions, misperceptions, or hallucinations--the object is not only, for example, a brick which is red and rectangular on the side facing me, it is seen as a brick which is red and rectangular on the facing side. How is this to be understood?

7. Traditionally a distinction was drawn between the visual object and the perceptual judgment about the object. The latter was construed as a special kind of occurrent believing. Occurrent acts of belief were, in their turn, construed as propositional in form; as having, so to speak, a syntactical form which parallels or is analogous to the syntactical form of the sentence which would express it in overt speech. Believings, so to speak, occur in Mentalese.

8. This suggested to some philosophers that to see a visual object as a brick with a red and rectangular facing surface consists in seeing the brick and believing it to be a brick with a red and rectangular facing surface:

This is a brick which has a red and rectangular facing surface

where the judgment has a demonstrative component analogous to the linguistic demonstrative, "this," in the sentence by which it would be expressed.

9. Now I think that there is something to the idea that seeing as involves an occurrent act of belief, but I also think that the standard account misconstrues the structure of the believing. Notice that the subject term of the judgment was exhibited above as a bare demonstrative, a sheer this, and that what the object is seen as was placed in explicitly predicative position, thus "is a brick which has a red and rectangular facing surface."

10. I submit, on the contrary, that correctly represented, a perceptual belief has the quite different form:

This brick with a red and rectangular facing surface

Notice that this is not a sentence but a complex demonstrative phrase. In other words, I suggest that in such a perceptually grounded judgment as

This brick with a red and rectangular facing side is too large for the job at hand

the perceptual belief proper is that tokening of a complex Mentalese demonstrative which is the grammatical subject of the judgment as a whole. This can be rephrased as a distinction between a perceptual taking and what is believed about what is taken. What is taken or, if I may so put it, believed in is represented by the complex demonstrative phrase; while that which is believed about the object is represented by the explicitly predicative phrase which follows. Perceptual takings, thus construed, provide the perceiver with perceptual subject-terms for judgments proper.

11. From this point of view, what the visual object is seen as is a matter of the content of the complex demonstrative Mentalese phrase.


12. I shall prepare the way for the next major step in the argument by changing my example. Consider the visual perception of a red apple. Apples are red on the outside (have a red skin) but white inside. Other features of apples are relevant, but this will do to begin with. The initial point to be made about the apple is that we see it not only as having a red surface but as white inside. This, however, is just the beginning. Notice that the experience contains an actual quantity of red. By "actual quantity of red" I mean a quantity of red which is not merely believed to exist as did the Fountain of Youth for Ponce de Leon. The Fountain of Youth does not actually exist. By contrast, the quantity of red which is a constituent of the visual experience of the apple not only actually exists but is actually or, to use a familiar metaphor, bodily present in the experience.

13. But what of the volume of white apple flesh which the apple is seen as containing? Many philosophers would be tempted to say that it is present in the experience merely by virtue of being believed in. It has, of course, actual existence as a constituent of the apple, but, they would insist, it is not present in its actuality. Phenomenologists have long insisted that this would be a mistake. As they see it, an actual volume of white is present in the experience in a way which parallels the red. We experience the red as containing the white.

14. But if what is experienced is red-containing-white as red-containing-white, and if both the red and the white are actualities actually present, how are we to account for the fact that there is a legitimate sense in which we don't see the inside of the apple? To be sure, we see the apple as white inside, but we don't see the whiteness of the inside of the apple.

15. We must add another distinction, this time between what we see and what we see of what we see. The point is a delicate one to which justice must be done if we are not to be derailed. Thus when I see a closed book, the following are all typically true.

(a) I see the book.
(b) The book has pages inside.
(c) I see the book as having pages inside.
(d) I do not see the inside pages.
(e) The book has a back cover.
(f) I see the book as having a back cover.
(g) I do not see the back cover.

16. How can a volume of white apple flesh be present as actuality in the visual experience if it is not seen? The answer should be obvious. It is present by virtue of being imagined. (Notice that to get where we have arrived, much more phenomenology must have been done than is explicitly being done on this occasion. We are drawing on a store of accumulated wisdom.)

17. But notice where this leads us. The actual volume of white is experienced as contained in the actual volume of red. Yet if the actuality of the white apple flesh consists in it being imagined, it must be dependent for its existence on the perceiver; it must, in a sense to be analyzed, be "in" the perceiver.

18. Before following up this point, it should be noticed that the same is true of the red of the other side of the apple. The apple is seen as having a red opposite side. Furthermore, the phenomenologist adds, the red of the opposite side is not merely believed in; it is bodily present in the experience. Like the white, not being seen, it is present in the experience by being imagined.

19. Notice that to say that it is present in the experience by virtue of being imagined is not to say that it is presented as imagined. The fruits of careful phenomenological description are not to be read from experience by one who runs. Red may present itself as red and white present itself as white; but sensations do not present themselves as sensations, nor images as images. Otherwise philosophy would be far easier than it is.

20. The phenomenologist now asks us to take into account a phenomenon frequently noted, but as frequently misinterpreted. Consider the snow seen on a distant mountain. It looks cool. Do we see the whiteness of the snow, but only believe in its coolth. Perhaps this is sometimes so; but surely not always. Sometimes actual coolth is present in the experience, as was the white inside the apple and the red on the opposite side. Once again, we do not see the coolth of the snow, but we see the snow as cool; and we experience the actual coolth as we experience the actual whiteness of the snow. An actual coolness is bodily present in the experience as is an actual volume of white.

21. Let us combine our results into one example. We see the cool red apple. We see it as red on the facing side, as red on the opposite side, and as containing a volume of cool white apple flesh. We do not see of the apple its opposite side, or its inside, or its internal whiteness, or its coolness, or its juiciness. But while these features are not seen, they are not merely believed in. These features are present in the object of perception as actualities. They are present by virtue of being imagined.

22. We must introduce a further refinement. We see an apple. We see it as an apple. Do we see of it its applehood? We see a copper penny. We see it as a copper penny. Do we see of it its consisting-of-copperness? We see a lump of sugar. We see it as white and as soluble. We see of it its whiteness. Do we see of it its solubility? The answer to the last question is surely negative, as are the questions concerning applehood and copperness, and for the same reason. Aristotle would put it by saying that we see of objects only their occurrent proper and common sensible features. We do not see of objects their causal properties, though we see them as having them.

23. To draw the proper consequences of this we must distinguish between imagining and imaging, just as we distinguish between perceiving and sensing. Indeed the distinction to be drawn is essentially the same in both cases. Roughly imagining is an intimate blend of imaging and conceptualization, whereas perceiving is an intimate blend of sensing and imaging and conceptualization. Thus, imagining a cool juicy red apple (as a cool juicy red apple) is a matter of (a) imaging a unified structure containing as aspects images of a volume of white, surrounded by red, and of mutually pervading volumes of juiciness and coolth, (b) conceptualizing this unified image-structure as a cool juicy red apple. Notice that the proper and common sensible features enter in both by virtue of being actual features of the image and by virtue of being items thought of or conceptualized. The applehood enters in only by virtue of being thought of (intentional in-existence). {1}

24. On the other hand, seeing a cool juicy red apple (as a cool juicy red apple) is a matter of (a) sensing-cum-imaging a unified structure containing as aspects images of a volume of white, a sensed half-apple shaped shell of red, and an image or a volume of juiciness pervaded by a volume of white; (b) conceptualizing this unified sense-image structure as a cool juicy red apple. Notice that the proper and common sensible features enter in both by virtue of being actual features of the sense image structure and by virtue of being items conceptualized and believed in. As before, the applehood enters in only by virtue of being thought of (believed in).


25. The upshot of the preceding section is that perceptual consciousness involves the constructing of sense-image models of external objects. This construction is the work of the imagination responding to the stimulation of the retina. From this point on I shall speak of these models as image-models, because although the distinction between vivid and less vivid features of the model is important, it is less important than (and subordinate to) the perspectival feature of the model (its structure as point-of-viewish and as involving containing and contained features).

26. The most significant fact is that the construction is a unified process guided by a combination of sensory input on the one hand and background beliefs, memories, and expectations on the other. The complex of abilities included in this process is what Kant calls the "productive" as contrasted with the "reproductive" imagination. The former, as we shall see, by virtue of its kinship with both sensibility and understanding unifies into one experiencing the distinctive contributions of these two faculties.

27. Notice once again that although the objects of which we are directly aware in perceptual consciousness are image-models, we are not aware of them as image-models. It is by phenomenological reflection (aided by what Quine calls scientific lore) that we arrive at this theoretical interpretation of perceptual consciousness.

28. Notice also that the construction of image-models of objects in the environment goes hand in hand with the construction of an image-model of the perceiver's body, i.e., what is constructed in an image-model of oneself-in-one's environment. The perspectival character of the image model is one of its most pervasive and distinctive features. It constitutes a compelling reason for the thesis of the transcendental ideality of the image-model world. Image-models are "phenomenal objects." Their esse is to be representatives or proxies. Their being is that of being complex patterns of sensory states constructed by the productive imagination.

29. Still more important is the fact that although the image-models are perspectival in character, the objects in terms of which they are conceptualized are not. Thus, apples are not perspectival in character. The concept of an apple is not the concept of a perspectival entity. Apples are seen from a point of view. Apples are imagined from a point of view. A spatial structure is imagined from a point of view. Yet the concept of a spatial structure, e.g., a pyramid, is not the concept of a point-of-viewish object. Thus we must distinguish carefully between objects, including oneself, as conceived by the productive imagination, on the one hand and the image-models constructed by the productive imagination, on the other.

30. We are now in a position to put the elements of visual perception which we have been distinguishing together.

31. In the first place, the productive imagination is a unique blend of a capacity to form images in accordance with a recipe, and a capacity to conceive of objects in a way which supplies the relevant recipes. Kant distinguished between the concept of a dog and the schema of a dog. The former together with the concept of a perceiver capable of changing his relation to his environment implies a family of recipes for constructing image models of perceiver-confronting-dog.

32. The best way to illustrate this is by a very simple example, for our perceptual experience does not begin with the perception of dogs and houses. The child does not yet have the resources for such experience. But though the child does not yet have the conceptual framework of dogs, houses, books, etc., he does, according to Kant, have an innate conceptual framework--a proto-theory, so to speak, of spatio-temporal physical objects capable of interacting with each other; objects--this is the crux of the matter--which are capable of generating visual inputs which vary in systematic ways with their relation to the body of a perceiver.

33. Consider the example of a perceiver who sees a pyramid and is walking around it, looking at it. The concept of a red pyramid standing in various relations to a perceiver entails a family of concepts pertaining to sequences of perspectival image-models of oneself-confronting-a-pyramid. This family can be called the schema of the concept of a pyramid.

34. Notice that the pyramid schema doesn't follow from the concept of a pyramid alone. It follows from the complex concept of pyramid in such-and-such relations to a perceiver. This accounts for the fact that whereas the concept of a pyramid is not a point-of-viewish concept, the associated schemas concern sequences of perspectival models of a pyramid.

35. It is in terms of these considerations that Kant's distinction{2} between (a) the concept of an object, (b) the schema of the concept, and (c) an image of the object, as well as his explication of the distinction between a house as object and the successive manifold in the apprehension of a house is to be understood. "The object is that in the appearance which contains the condition of this necessary rule of apprehension."{3}

36. To sum up, the productive imagination generates both the complex demonstrative conceptualization

This red pyramid facing me edgewise

and the simultaneous{4} image-model, which is a point-of-viewish image of oneself confronting a red pyramid facing one edgewise. We are now in a position to understand Kant's distinction between the productive and the reproductive imaginations.{5} The principle of the reproductive imagination is the "association of ideas"; more exactly, the association of objects. The connection between the associated items is contingent, and dependent on the happenstances of experience. As an association of objects it presupposes the constitution of objects by the productive imagination. And the principle of such constitution is not happenstance, but conformity to the recipe--schemata derived from concepts. We are also in a position to understand the precise sense in which the productive imagination mediates between "the two extremes, . . . sensibility and understanding . . ." (A124) and is ". . . an action of the understanding on the sensibility" (B152).


37. In the preceding section of the paper I emphasized a distinction between what we perceive the object as and what we perceive of the object. I related this distinction to the distinction between the complex demonstrative thought component and the complex-image component of the perceptual experience. I now want to do this in a more systematic way and to relate it to Kant's theory of categories.

38. The basic idea Is that what we perceive of the object in visual perception consists of those features which actually belong to the image-model, i.e., its proper and common-sensible qualities and relations. Also its perspectival structure. On the other hand, what we perceive the object as is a matter of the conceptual content of the complex demonstrative thought. I pointed out that the sensible features belong in both contexts. Thus the phrase "cube of pink (from a certain point of view)" refers both to an actual feature of the image-model and (in second intention) a component of the conceptual center of the demonstrative thought. Thus,

39. Now I emphasized that we do not perceive of the object its causal properties. What we see of it are its occurrent sensible feature. This can now be generalized as follows. We do not perceive of the object its character as a substance having attributes, its character as belonging with other substances in a system of interacting substances, its character as conforming to laws of nature. In short, we do not perceive of the object what might be called "categorial" features. For the image construct does not have categorial features. It has an empirical structure which we can specify by using words which stand for perceptible qualities and relations. But it does not have logical structure; not-ness, or-ness, all-ness, some-ness are not features of the image-model. They are feature of judgment. More generally we can say that the image-model does not have grammatical structure. (It will be remembered that we are construing mental judgments as analogous to sentences. A judgment, we said, is, as it were, a Mentalese sentence episode. And, of course, Kant's categories are grammatical classifications. They classify the grammatical structures and functions of Mentalese.

40. Thus the category of substance-attribute is the structure 'S is P', the form of subject-attribute judgment. The category of causality is the form 'X implies Y'. The category of actuality is the form 'that-p is true'. More accurately, the categories are these forms or functions specialized to thought about spatio-temporal object.

41. In the preceding section we were concerned with the distinction between concepts of empirical object and the schemata of these concepts, i.e., the rules for image-model sequences which are determined by the concept of object-in-various-successive-relations-to-perceiver. But Kant also uses the term schema in connection with the categories. The categories do not specify image-models. There is no image of causality as there is an image of a house. Yet they do have in their own way schemata, i.e., rules specified in terms of abstract concepts pertaining to perceptible features of the world. Thus the schema for causality is the concept of uniform sequence throughout all space and time.

42. The Humean concept could be said to have images in an extended sense. Thus a person in a thunderstorm who experiences a finite stretch of lightning-thunder uniformity could be said to have experiences an image of causality. Kant, of course, does not say this, and I introduce it only to show that this new use of "schema" is not completely foreign to the previous one.

43. The schematized category of causality, then, is the ground-consequence category where the ground (antecedent) concerns the occurrence of one kind of event, K1, at t and the consequence concerns the occurrence of another kind event, K2, at t + (delta t). Since the ground is an event being of kind K1 it must be true that whenever K1 occurs, K2 also occurs.

44. The categories are in first instance simply identical with the forms of judgment, a point which must be grasped if traditional puzzles about the metaphysical deduction of the categories are to be avoided. These forms of thought would be involved in thinking about any subject matter from perceptual objects to metaphysics and mathematics.

45. The so-called pure categories are these forms of thought specialized to thought about objects (matter-of-factual systems) in general. Such objects need not be spatio-temporal, as are the objects of human experience. The full-blooded categories with which Kant is concerned in the Critique are the pure categories, specialized in their turn to thought about spatio-temporal objects. The relation of the forms of thought to the pure categories is that of genera to species, as is the relation of the pure categories to schematized categories.


46. Let me conclude with some remarks on Kant's concept of an intuition. Consider the statements

This is a pyramid
This pyramid is made of stone

The first has the explicit grammatical form of a sentence. So does the second. But notice that the grammatical form of a sentence is lurking in the subject of the second sentence. From the standpoint of transformational grammar we would think of it as derived from the deep structure.

This is a pyramid and it is made of stone

47. One might be tempted to think of "this" as a pure demonstrative having no other conceptual content than that involved in being a demonstrative. Kant does think of an act of intuition as a demonstrative thought, a Mentalese "this." However he does not think of this Mentalese demonstrative as a bare Mentalese "this." An example of an act of intuition would be the Mentalese counterpart of

This cube facing me edgewise.

where this is not to be understood as, so to speak, a Mentalese paraphrase of

This is a cube which faces me edgewise.

48. The role of an intuition is a basic and important one. It is the role of bringing a particular object before the mind for its consideration. Thus, though there is a close relationship between

This cube facing me edgewise . . .


This is a cube which faces me edgewise.

the former is an irreducible kind of representation. It is a demonstrative representation which has conceptual content and grammatical form. As noted above it contains the form and content of the judgment "This is a cube."

49. Thus for Kant intuitions are complex demonstrative thoughts which have implicit grammatical (and hence categorial) form.

50. However thin--as in the case of the child--the intuitive representation may be from the standpoint of the empirical concept involved, it nevertheless contains in embryo the concept of a physical object now, over there, interacting with other objects in a system which includes me. It embodies a proto-theory of a world which contains perceivers of objects in that world. (The reader should ponder A127-8.)

51. Kant emphasizes the difference between intuitions on the one hand and sensations and images on the other. He emphasizes that it is intuitions and not sensations or images which contain categorial form. When he speaks of synthesis in connection with perception, he has two things in mind:

(1) the construction of image-models
(2) the formation of intuitive representations (complex demonstratives)

There is also the synthesis which is the formation of the explicit judgment, thus

(3) (This cubical substance) is a piece of ice

52. Since intuitions have categorial form, we can find categorial form in them. In this sense we can arrive at categorial concepts by abstracting from experience--but only because experience contains intuitions which have categorical form. We cannot abstract the categories from sensations or images.

53. To sum up. Kant's categories are forms and functions of judgment. They are grammatical summa genera. From Kant's point of view, Aristotle's theory of categories was a failure because, failing to distinguish between intuition and image-model, his list of categories is haphazard (though guided obscurely by grammatical intuitions); and Aristotle confuses them with generic concepts of entities in the world. There is a legitimate place for a theory of such concepts. But it must be carefully distinguished from the grammar of thought.

[Return to Main Page]


{1} The essential point is that the experience involves the actual rather than merely believed in existence of the volume of white. The phenomenologist thinks of this volume as an image-volume of white. A tough-minded contemporary might insist that while the actuality is a three-dimensional neuro-physiological process of the kind correlated with stimulation by white objects, it need not be construed as an image in a more traditional sense. Nothing in the argument of this paper hinges on this controversy. One can think of the neuro-physiological process as occasionally generating image-samples of white. On the other hand, when we look, not at an opaque object but, for example, at a pink ice cube, the neuro-physiological process involves the actual existence of a volume of sensory pink. [Back]

{2} A140ff.; B179ff. That Kant does not emphasize the role of the concept of the perceiver's body and sense organs in determining the rules of the successive apprehension of the manifold of an object (e.g., a house) is, presumably, to be accounted for by the fact that the concept of the perceiving subject and its faculties is a constant factor in the generation of all object-schemata. The source of the difference between different object-schemata is the difference between the object-concepts involved (e.g., house, dog, etc.). The reader should ponder Kant's examples of the house and the ship in explaining the distinction between subjective and objective succession (A192; B237). [Back]

{3} A191; B236. For a more detailed explication of this Kantian thesis, which plays a central role in the transcendental deduction--most explicitly in the first edition (A104, A120)--see my paper, "Kant's Transcendental Idealism," Proceedings of the International Kant Congress, Ottawa, 1976. [Back]

{4} A momentary image-model is, of course, an abstraction from the image-model sequence which realized the schema for a particular sequence of relations between perceiver and object. [Back]

{5} See A120-4. The discussion of the role of the productive imagination in the second edition transcendental deduction is very abstract and emphasizes the self-consciousness involved in the construction and awareness of perceptual representations (but see B138; B150ff.). The specifics were left to the Analytic of Principles, where the account coincides with the first edition.[Back]

[Return to Main Page]