Wilfrid Sellars, "Some Reflections on Perceptual Consciousness," in Selected Studies in Phenomenology and Existential Philosophy, edited by R. Bruzina and B. Wilshire (The Hague, 1977): 169-185.


1. My aim in this paper is to put together some views on visual perception which I have developed in rather fragmentary form -- usually subordinated to other topics -- since my first attempt at a synoptic view in "Empiricism and the Philosophy of Mind."1

2. I still think that the general strategy of that essay was sound; but its explicit results were so schematic that, in spite of the fact that one of its central themes was a reassessment of the treatment of visual perception by philosophers in the analytic tradition, it provides nothing worthy of being called a theory of visual perception. Nevertheless such a theory was obviously hovering in the background, if only as a research program and an attempt to specify certain adequacy conditions which such a theory would have to satisfy.

3. In the allotted time I can scarcely present a full fledged theory of visual perception. I can, however, turn my attention directly to some of the problems with which it deals, instead of subordinating them to other topics, let alone to an attempt to sketch an entire philosophy of mind. The situation will call for a delicate balance of assumptions and arguments, of raising hares and letting sleeping dogs lie.

4. Having thus construed my task, I do not expect my remarks to contain many surprises. Perception theory has been developed by able men, and the alternatives they present can be expected not only to contain a large measure of truth, but to agree in ways which, though not easily noticed because of differences in the larger contexts in which they are embedded, are nevertheless genuine.

5. Furthermore, for longer than I care to remember I have conceived of philosophical analysis (and synthesis) as akin to phenomenology. I would therefore expect this audience to be more sympathetic to what I have to say than many of my colleagues would expect. On the other hand, since I shall be dealing with specifics, I also would expect this sympathy to be laced with disagreement. Indeed some measure of disagreement is exactly what I am hoping for. After all, disagreement presupposes communication.


6. The primary datum to be approached from the standpoint of conceptual analysis (or phenomenology) is that in the standard or paradigm cases we see physical objects from a point of view in physical space; thus, a red brick over there facing me edgewise

7. Phenomenological reduction begins when we distinguish between the object seen (in this case an opaque object) and what we see of the object. We see the brick, but of the brick we see certain of its surfaces. These surfaces are, in a sense which I shall not attempt to define, dependent particulars. If one is careful, one can call them 'parts' or 'constituents' of the brick. They, too, we see from a point of view in physical space.

8. It is important that we think of these perceptible constituents as particulars. For what we see of an object also includes, as will shortly emerge, certain universals (attributes, relational properties); and universals are in no useful sense constituents of objects.

9. It is customary to distinguish between

seeing a physical object (e.g. a brick)
seeing that the physical object is a brick
seeing that the physical object has a red facing surface.
Taking into account the above distinction between what we see (e.g. a brick) and what we see of what we see (e.g. a certain part of its surface), one would add a distinction between
seeing of a physical object its facing surface
seeing that the facing surface of a physical object is (e.g.) red.

10. Schematically the distinction is between seeing objects, thus

seeing O (where O is a physical object)
seeing O' (where O' is a perceptible constituent of O)
and what is often called 'propositional seeing', thus
seeing that O (or O') is φ

11. These distinctions are reflected in traditional accounts of the mental activity involved in visual perception. I shall limit my remarks to those accounts which speak of perceptual takings, where takings are construed as occurrent beliefs (in the sense of believings) which (given ones perceptual set) we are caused to have in visual perceptual situations. In standard conditions the objects seen are in a legitimate sense (by no means easy to analyze) the external causes of takings.

12. Occurrent believings are construed as mental acts, the appropriate expression of which is the tokening of a sentence. Where the sentence is a subject-predicate sentence, we can speak of the subject and the predicate of the corresponding believing. It is commonly held that perceptual takings have subject-predicate form and that the subject constituent is appropriately expressed by a demonstrative. This constituent is itself construed as a demonstrative, not because acts of belief are linguistic, but because they are sufficiently analogous in essential structure and function to the sentence tokens which express them in candid speech, for it to be appropriate to make an analogical use of semantical terminology in describing them.

13. It is often noted that we express at least some of our perceptual experiences by using sentences, the grammatical subject of which is a complex demonstrative phrase, thus

This red brick is larger than that one
This sentence is obviously related to the compound sentence
This is a brick and it is red and it is larger than that one
Those who carefully distinguish between thought and its verbal expression can be tempted to construe the thought which the first of these sentences expresses in terms of the structure of the second sentence.

14. If this move is made, the result is to construe the subject of the perceptual taking as a bare 'this,' all characterizing being put in an explicitly predicative position; thus, to return to our original example, This is a red brick facing me edgewise.

15. The referent of 'this' in perceptual contexts is construed as the object seen. Thus, if the referent is a certain black bush, the perceiver is said to see the bush. If his taking has the form

This is a bear
then he is said to take what is in point of fact a bush to be a bear. In our example, which I shall suppose to be a case of veridical perception, the perceiver sees a certain red brick facing him edgewise and sees that it is a red brick facing him edgewise, by virtue of the occurrence of a perceptual believing or taking of the form
This is a red brick facing me edgewise.

16. But this division of visual takings into a subject which is a pure demonstrative and predicative constituent in an explicitly predicative position simply won't do as it stands. To appreciate this, however, we must back up a little.

17. Thus it is not clear, to begin with, that all perceptual takings are to be construed on the model of sentences with demonstrative subjects, though they may (and, perhaps;, must) contain a demonstrative component. Thus, on certain occasions one can correctly be said to see that an airplane is flying high overhead without seeing the airplane. Perhaps in such cases the taking has the form

There is an airplane flying way up there
On the other hand, a distinction between the situation we see, and what we see of the situation may convince us of the primacy, in some sense, of the demonstrative analysis. What we see of the situation might be, for example, a vapor trail which grows at one end and fades at the other. The vapor trail is an object, and it seems natural to generalize the example and conclude that visual perception always in some sense includes the perception of an object,2 and a taking of the form, to use the above example,
This is a vapor trail

18. Leaving this issue aside for the moment, it must also be noted that even within the demonstrative mode1 a distinction must be drawn between what we see, and what we see it as. The point is a familiar and important one, and part of its importance lies in the fact that the fact that the pure demonstrative model is incapable of handling it.

19. On the other hand, the complex demonstrative phrase model shows promise. Thus it is not implausible to suggest that one who sees a bush as a bear has a perceptual belief or taking of which the subject constituent is the complex demonstrative

This large black bear ...

20. If the believing of which the demonstrative phrase is the subject can be said to be a perceptual believing, perhaps

This large black bear is moving toward me
then we might distinguish between
takings that
takings as
and suggest that what one sees an object as is what belongs in the demonstrative phrase along with 'this.'

21. Indeed, one might go further and say that, properly speaking, takings simply are the complex demonstrative constituents of perceptual beliefs and that the explicitly predicative constituent of the belief is not part of what is taken, but simply what is believed about what is taken.

22. The model for taking, then, would be presupposition in something like Strawson's sense. The concept of occurrent belief could be extended to cover this sense of taking, by distinguishing between believing that and believing in. A perceptual believing in would be illustrated by the subject constituent of the believing expressed by

This red brick facing me edgewise is too large to fit that gap.

23. The complex demonstrative constituent could be construed as a presupposing that the referent of 'this' is a red brick facing one edgewise, and argued that by virtue of this presupposing the referent of 'this' is perceptually taken by the perceiver as a red brick facing him himself edgewise.

24. Suppose that the referent of 'this' is a brick shaped piece of red modeling clay. Is the occurrence of the taking sufficient to warrant us in saying that the perceiver (Jones) sees a red brick shaped piece of red modeling clay facing him edgewise as a red brick facing him edgewise?

25. Clearly part of the problem is to give a clear account of the sense in which the 'this' of

This red brick facing me edgewise ....
can be construed as having a reference which is independent of the predicates which accompany it in the complex demonstrative phrase, so that it makes sense to say that 'this' refers to something which is not a red brick. The proper move to make here seems to be the move from the phrase to what it presupposes, put in explicitly propositional form, e.g.,
This is a red brick,
and determine the referent of 'this' in the latter context. One must be careful to do this, however, without construing the perceptual believing expressed by
This red brick facing me edgewise is too large
as identical with the perceptual believing expressed by
This is a red brick and it faces me edgewise and is too large.
These considerations will loom larger at a later stage of the argument.

26. If we leave them aside, the question raised above (in paragraph 24) can be put more accurately as follows: Should we say that Jonas' seeing a certain object as a red brick facing him edgewise consists in his believing in a red brick facing him edgewise, where this believing in is visual in the sense that his having this belief is, given his mental set, brought about by the action of that object on his visual apparatus?


27. Before tackling this question we must refine our distinction between the object seen and what we see of the object. For what we see of an object includes not just the dependent particulars we call 'parts' or 'constituents' (e.g. the surface of the brick). It also, as was suggested at the time, includes certain qualities or attributes and relations. That it includes them will be clear enough, though just how it includes them will be highly problematic.

28. Consider, for example, my favorite object, this pink ice cube. Its consisting of ice is largely a matter of its causal properties. It cools tea and is melted by fire.

29. We see the pink ice cube and, supposing the conditions of perception to be normal, we see that the transparent pink cube is made of ice. We even see the pink ice cube as a cube of pink ice. But do we see of the pink ice cube the causal properties involved in its being made of ice?

30. Of course in seeing it as a pink ice cube we are seeing it as having the causal properties characteristic of ice. But do we see of the object these causal properties? The question is, to be sure, an awkward one. But if we consider certain other questions of a similar form, the answer seems to be no.

31. Thus, by contrast, we not only see that the ice cube is pink, and see it as pink, we see the very pinkness of the object; also its very shape -- though from a certain point of view.

32. Even more interesting is the fact that in seeing the cube as a cube of ice we are seeing it as cool. But do we see of the cube its coolness? Here we are torn in a familiar manner. On the one hand we want to say that the pinkness and coolth are 'phenomenologically speaking' on a par, and are tempted to say that the idea that we don't see its coolness is a matter not of phenomenology, but of scientific theory. On the other hand, when asked point blank whether we see its coolness, we find an affirmative answer intuitively implausible and are tempted to fall back on the idea that its coolness is believed in, i.e. that it is taken as cool.

33. Obviously we want to say that the ice cube's very coolness is not merely believed in, even though its very coolness is not seen. It clearly won't do to say that we feel or imagine a coolness on seeing the cube; what is in question is its coolness.

34. But though this topic is of great intrinsic importance, it must be postponed to another occasion, although I believe that the framework I am about to develop -- enriched with an account of synaesthesia -- provides the key to the answer.


35. Sufficient to the occasion is an analysis of the sense in which we see of the pink ice cube its very pinkness. Here, I believe, sheer phenomenology or conceptual analysis takes us part of the way, but finally lets us down. How far does it take us? Only to the point of assuring us that something, somehow a cube of pink in physical space is present in the perception other than as merely believed in.

36. In traditional terminology, the somehow presence of a cube of pink does not consist in its intensional in-existence as the content of a conceptual act. Nor is its character as a cube of pink in physical space facing me edgewise a matter of its actually being a cube of pink in physical space. It is somehow a cube of pink in physical space facing me edgewise without actually being a cube of pink in physical space facing me edgewise, yet without merely being the content of a belief in a cube of pink in physical space facing me edgewise.

37. Seeing of the cube its very pinkness and its cubicity (from a point of view) would be analyzed in terms of this somehow, other than merely believed in presence of a cube of pink in physical space facing one edgewise in the visual experience.

38. I say 'visual experience' because it is time to take into account, at least provisionally, the fact that we can seem to see a cube of pink ice from a point of view in physical space when there is, in point of fact, no cube of pink ice in the neighborhood.

39. We can use the phrase 'ostensible seeing of a cube of pink ice facing one edgewise as a cube of pink ice facing one edgewise' to refer to a visual experience which would be a case of seeing a cube of pink ice facing one edgewise as a cube of pink ice facing one edgewise, if there was such a cube of pink ice and it was (in a sense requiring analysis) causally responsible for the ostensible seeing.

40. Thus the somehow, other than as believed in, presence of a cube of pink ice facing one edgewise would be common to what can provisionally be called veridical and non-veridical ostensible seeings of a cube of pink ice facing one edgewise.

4l. Now the obvious move is to introduce visual sensations as proto-theoretical states of perceivers to explain these results of phenomenological or conceptual analysis. Thus the fact that we see (veridical perception) from a point of view or merely ostensibly see (unveridical perception) from a point of view, the very pinkness and cubicity of a cube of pink ice would be explained by postulating the occurrence in the perceivers of a sensation of a cube of pink facing one edgewise. The fact that we don't see its very character of being made of ice would be explained by the exclusion of the proto-theory of the expression 'sensation of a cube of ice'.

42. It is by the introduction of visual sensations that we transcend phenomenology or conceptual analysis. They are not yielded by phenomenological reduction, but postulated by a proto-(scientific)- theory.

43. A sensation of a cube of pink facing one edgewise is a sensation of a certain kind, the kind normally brought about by the action on the visual apparatus of the perceiver by transparent physical cubes of pink (e.g. pink ice) which face the perceiver edgewise from a certain distance.

44. Visual sensations, which are states of the perceiver, are not (for example) literally cubes of pink facing the perceiver edgewise. On the other hand it is not simply false that they are cubes of pink facing the perceiver edgewise. It is tempting to appeal to the tradition of analogy and say that the pinkness and cubicity of a sensation which belongs to the 'of a cube of pink' kind are analogous to the pinkness and cubicity of its standard cause.

45. But analogies are useful only if they can be cashed or spelled out. One way of doing this is by saying -- as I have said on a number of occasions --

According to this version of the adverbial theory of sensing, then, sensing a-pink-cube-ly is sensing in a way which is normally brought about by the physical presence to the senses of a pink and cubical object, but which can also be brought about in abnormal circumstances by objects that are neither pink nor cubical, and, finally, according to this form of the adverbial theory, the manners of sensing are analogous to the common and proper sensibles in that they have a common conceptual structure. Thus, the color manners of sensing form a family of incompatibles, where the incompatibilities involved are to be understood in terms of the incompatibilities involved in the family or ordinary physical color attributes. And, correspondingly, the shape manners of sensing would exhibit, as do physical shapes, the abstract structure of a pure geometrical system.3

46. I have come to see, however, that we must be able so to formulate the analogy between manners of sensing and perceptual attributes of physical objects, that it is made evident that the analogy preserves in a strict sense the conceptual content of predicates pertaining to the perceptible attributes of physical objects, while transposing this content into the radically different categorial framework to which manners of sensings belong. Just what more (if anything) this would involve than spelling out in greater detail the analogies referred to above, and (perhaps) adding additional dimensions of analogy, I am not able to say. I believe, however, the problem is an important one, and that an adequate answer is necessary to explain the sense in which color concepts preserve their content throughout their migration from the manifest image to the scientific image.

47. In particular, the idea that there is a sense in which conceptual content can be preserved through a change of category seems to me necessary to give meaning to the idea that the very pinkness and cubicity of pink ice cubes can be somehow present in ostensible seeings of pink ice cubes as pink ice cubes.


48. It is an essential feature of the sensings postulated by the proto-theory I have been describing that they are not sensing as. To sense (a cube of pink)ly is not to sense something as a cube of pink, though it is a state postulated by a theory designed to explain what it is to see (or seem to see) a cube of pink as a cube of pink. Thus, sensing, though it is a constituent of seeing something as something, is not itself a case of seeing something as something.

49. I have distinguished two constituents of an ostensible seeing of a cube of pink ice as a cube of pink ice,4

  1. the taking or believing in, construed on the model of the complex demonstrative phrase
    This cube of pink ice

  2. the sensing (a cube of pink)ly.
The crucial question concerns how these constituents are related, the kind of togetherness they have.

50. It might be thought that the relation is merely a causal one, i.e. that given a certain perceptual set, the sensing is the immediate cause of the taking. If the perceptual set includes the belief that the room is illuminated by pink light, the perceiver might have a perceptual belief in a colorless ice cube. In this case he would be conceptualizing a cube of colorless ice, though sensing (a cube of pink)ly.

51. Would we say that he is seeing a cube of pink ice as a cube of colorless ice? In any event, he would not be seeing its colorlessness. A necessary condition of seeing its colorlessness is to sense (a cube without color)ly.

52. That the relation between the sensing and the taking is at least in part that of the former (given a certain perceptual set) the immediate cause of the latter, is, I believe, clear. Might not the relation be even more intimate?

53. To appreciate this possibility, let us ask: What, in a case of veridical perception, is the referent of the demonstrative phrase

This cube of pink ice ...
The obvious answer would seem to be a certain cube of pink ice.

54. Suppose that what is in fact there is a cube of colorless ice. Would we say that the reference has failed, i.e. that the demonstrative phrase has no referent? Perhaps we would be willing to say that it does have a referent, namely, the cube of colorless ice, but that the perceiver is taking it to be a cube of pink ice. Would we also say that he sees the referent as a cube of pink ice? We can not say that he sees its very pinkness, but we can say that he ostensibly sees (or seems to see) its very pinkness. A similar move can be made if we suppose that what confronts the perceiver is a cube of glass.

55. But suppose that there is nothing there, i.e. that the perceiver is hallucinating. Would we say that the reference has failed, i.e. that the demonstrative phrase has no referent? Many philosophers would say that the answer is obviously yes. They would argue the previous move to save reference is no longer available for there is nothing over there in physical space of which it can be said that the perceiver takes it to be a cube of pink ice. The latter is, indeed, the case ex hypothesi. But is it the end of the matter?

56. What of the possibility that when all of the presuppositions packed into the complex (Mentalese) demonstrative phrase

This cube of pink ice facing me edgewise ...
have been put into explicitly propositional form, it makes sense to preserve the reference of 'this' by construing the referent as the sensation of a cube of pink facing one edgewise?

57. This suggestion faces a high hurdle, however, in the fact that if we construe the process of making the presuppositions embedded in the (Mentalese) demonstrative phrase explicit on the model of moving in the direction of minimal perceptual takings, e.g.

This dangerous black bear standing on its hind legs ....
This dangerous black bear ....
This black bear ....
This black object ....

the minimal takings belonging to such series seem to concern items in physical space, thus, as in our example,
This cube of pink (over there) facing me edgewise.
The latter seems to require that the referent of 'this' must be the sort of thing which could be over there facing me edgewise. And it is surely a categorial feature of sensations that they are not over there facing me.

58. Nevertheless the possibility remains that whereas one can properly deny that the perceiver is seeing anything over there in space facing him edgewise, let alone a cube of pink ice, it would be incorrect to deny that he sees anything. After all, the perceiver is not imagining something, which is what the denial that he is seeing anything would normally imply.

59. Indeed, if we take seriously the idea that the thinning out of perceptual commitment which is implied by phenomenological reduction ends not with

This cube of pink over there facing me edgewise ....
but rather with
This somehow (a cube of pink over there facing me edgewise) ...
then the way would be open to save the reference by construing it to be the sensation, for the sensation is indeed that in the experience which is somehow a cube of pink over there facing one edgewise.

60. Notice that this move does not require the perceiver to conceptualize his sensation as a sensation.

61. Notice also that 'somehow' admits as a special case 'straightforwardly.' Thus while the referent of the most cautious perceptual taking can be construed as a sensation, we need not conclude that the referent of all perceptual takings is a sensation. For while it could be argued that the ultimate referent is always a sensation, by construing our original complex demonstrative phrase along the lines of phenomenological reduction as

This somehow (a cube of pink facing me edgewise) which is a cube of pink ice facing me edgewise ....
the initial stages of reference saving can proceed without interpreting the referent as a sensation.

Wilfrid Sellars

Pittsburgh, October 15, 1975


1 Minnesota Studies in the Philosophy of Science, Vol. I, edited by Herbert Feigl and Michael Scriven, pp. 253-330; reprinted as Chapter 5 of Science, Perception and Reality, London, 1965.

2 Of course the object need not be a physical object in the narrow sense in which the red brick is an example. It might be a flash of lightning.

3 "The Structure of Knowledge: Perception", the first of a series three Matchette Lectures for 1971 at the University of Texas, published in Action, Knowledge and Reality, edited by Hector-Neri Castaneda, Bobbs-Merrill, Indianapolis, 1975, p. 313.

4 For purpose of, simplification, I omit the perspectival element.