Romane Clark, "Sensibility and Understanding: The Given of Wilfrid Sellars," Monist 65 (1983): 350-364.

SENSIBILITY AND UNDERSTANDING:
THE GIVEN OF WILFRID SELLARS

Romane Clark
Indiana University

41. But is it genuinely necessary to interpose nonconceptual representations as states of consciousness between the 'physical' impact of the sensory stimulus and the conceptual representations (guarded or daring) which find verbal expression, actually or potentially, in perceptual statements? Can we not interpret the receptivity involved in terms of 'purely physical' states, and attribute to these the role of guiding conceptualization? Why should we suppose that receptivity culminates in a state which is neither 'purely physical' nor conceptual?1

Why indeed? Yet the postulation of just such nonconceptual states of consciousness is crucial to Professor Sellars's account of the way in which sensing is essential to perceiving.

Paragraph 41 is the lead paragraph of Sellars's Science and Metaphysics, Chapter I, "Sensibility and Understanding," Part IV. The non-conceptual states of consciousness to which he refers in these first three sentences of that lead paragraph he calls "sense impressions." Sellars contrasts sense impressions both with perceptions in general and with certain perceptual references which he calls "minimal perceptual takings" in particular. We think here (tendentiously) of Sellars's sense impressions as Sellars's sense-data. They are his "given", they are the (postulated) nonconceptual items of direct awareness upon which all perceptions and ostensible perceptions are based.

In what follows I state what I take Sellars's account of sense-impressions to be. This involves specifying how they differ from, but also are related to, the minimal perceptual takings and acts of perception with which they are contrasted. Given our characterization of these various elements figuring in Sellars's account of perception, it is then possible to return to the initial question of Paragraph 41: We ask whether it is indeed necessary to postulate the existence of sense impressions in Sellars's sense; whether it is necessary to interpose occurrences of them between the physical and conceptual items otherwise necessary for the occurrence of a perceptual act.

1. The Elements of Perception.

Sellars2 holds that acts of perception are complex acts. Perceiving, as well as ostensibly perceiving, what passes "before" one is rather like thinking or judging what is the case. Propositional perceptions, our perceptions that things or happenings in our surroundings are thus and so, are complex in a manner similar to the way in which judgments or thoughts are complex. Just as, in our logically simplest thoughts, there is both an object of reference and an attribution to that which has been thought of, so too in the simplest perceptions there is that which is perceived and an awareness of how or what that which is perceived happens manifestly to be. Given this minimal sort of logical complexity which is common to the propositional form of simple judgments and perceptions, as we record these, it is plausible to suppose that there exists a matching sort of complexity in the very acts of judging or thinking or perceiving what is the case. Accordingly, thinking and perceiving what is the case are, on this assumption, acts which consist of, and acts which are transacted in the occurrences of, certain constituent and perhaps even constituting acts of reference and attribution.

Supposing that perceiving consists of constituent acts of reference, and attribution, does it consist wholly of these? It might for instance be the case that it does not. It might be that some further, supervenient act is required which must occur and operate in some appropriate way upon these constituent acts to yield a perception. This is, perhaps, a possible view, but it is not, I believe, Sellars's view. There is no further, supervenient act required to tie reference and attribution together on Sellars's account. Certainly, there is none in the case of simple, thoughts. It is even less likely for perceiving given the seamless nature of perceptual occurrences.

This is not to say, however, that constituent mental occurrences of referential and attributive acts fully constitute an act of perception. Even if such acts constitute thoughts in (elementary acts of) nonsensuous thinking, it is not true of perceiving. There is in perceiving, always, a certain further complexity, not present in nonsensuous thinking. Perceiving what is the case, however similar in certain ways to thinking what is the case, differs radically from it as well. This further complexity explains, in fact, the palpable differences between. nonsensuous thinking and perceiving.

The palpable differences between thinking and perceiving are many, and dramatic. We can after all see something, we know not what. We may then look again, or look more closely; we may puzzle and speculate over what it may be which we perceive. But we cannot think of something, we know not what. Our puzzlings and speculations concerning objects of thought guarantee, because they presuppose, articulate reference. I cannot wonder whether it is John I think of in the way I can wonder whether it is John I see.

Further, there are the dramatic facts that all perceiving is perspectival and all is (at least tacitly) demonstrative. Perception is necessarily linked to the sensorily available surroundings on the occasions of its occurrences. But thought of course is not in this way restricted. You may stand up, thereby blocking my view. You keep me from then seeing, but not thinking, what is the case. On the other hand, I can of course think of, but not perceive, what is past, sensorily remote, too fast or too tiny to be sense. Nonsensuous thinking, but not perceiving, can outrun the boundaries of the occasions of its occurrences.

There is, then, in the twin facts of how perceptions are necessarily linked to the occasions of their occurrences, whereas thoughts need not be, and how acts of perception are essentially sensuous in character, the basis for a further complexity necessary to acts of perceiving which is lacking in acts of thought in general. Given these twin facts, it is natural to locate this further complexity in the role and nature of sensing. There is, after all, no perceiving without sensing. But how then do acts of sensing figure in acts of perceiving how things are? Are they literally constituents of perceptions? Or are they, rather, just occasions with the potentiality for acts of perception; causal triggers for perceiving? Evidently, acts of sensing enter somehow into the complex set of conditions necessary for perception. Do they literally enter into perception itself?

Sellars holds that acts of sensing are constituent acts not merely concomitant to but partly constituting perceivings. They are not merely causal triggers, the stimuli for perceptions, although they are also that. They enter perception in a more intimate way. Acts of sensing together with certain conceptual acts of demonstrative reference and attribution, are co-constituents in, and with them constitute, acts of perception. It is by embodying these constituent acts of sensing that perceptions gain their demonstrative character, the element which ties them to their sensory environment on the occasion of their occurrence. It is through these sensings that perceptions take on their palpable sensuous content. It is these constituent acts of sensing that, in Sellars's phrase, guide the conceptualizations which in turn determine which particular acts of reference and ascription occur in a given, specific, act of perception.

Acts of sensing are themselves impressions of sense. They are, e.g., visual sensations of color or tactual experiences of texture. They are impressions of the proper and common sensibles of Aristotle, those instances of the qualities and relations given to, and obtained by, the senses.

It needs to be stressed that these sense impressions, these acts of sensing, are not on Sellars's account themselves acts of awareness. Although they are mental occurrences, acts of consciousness, there is nothing intentional or conceptual about them in their own right. Quite to the contrary, it is Sellars's point that these enter intimately into perceptions, on the occasions they do so, not as conceptualizations of what we experience, but as themselves objects of conceptual awareness. It is to them that our acts of reference, our perceptual demonstrations, tacitly or explicitly (in minimal perceptual takings) make reference. It is Sellars's startling thesis that our perceptions, all of them, are not in reality about physical things and happenings in our environment but about certain mental events in our heads. These mental events turn out of course to be the impressions of sense. These mental events are acts of sensing.

Sellars is an indirect, representative realist. To the traditional philosophical question: What is it of which we are directly and noninferentially aware in perception? he answers: Sense impressions. To the question: How do our perceptions relate to the physical objects and happenings they seem to be about? he offers a subtle characterization of the nature of sense impressions and how they figure in perception. I summarize now what seem to me to be essential theses included or implied in those which Sellars postulates of sense impressions.

1. Sense impressions are sets of states of consciousness. They are constituted by the sequences of mental states which collectively comprise an act of sensing, from its onset to termination.

2. Sense impressions are acts of sensing.

3. Although mental occurrences, sense impressions are not intentional or cognitive. Occurrences of sense impressions neither involve nor presuppose any conceptual capacities or potentialities of those who suffer these occurrences.

4. Sense impressions are not acts of awareness. A sense impression of a red cube, even in the circumstance that there is before one a cube whose presence is a part-cause of the occurrent act of sensing, is not an awareness of a cube or, for that matter, of anything.

5. The occurrence of sense impressions does not validly justify what. Chisholm called "the sense-datum inference."3 I.e., it does not follow from the fact that one has a sense impression of red, that there is anything red which one senses.

6. Sense impressions as acts of sensing have of course their own properties. On the one hand, as events or perhaps processes, sense impressions have generic characteristics which determine the sort of event or process they are: a sensing, but not an act of willing; a mental event, not a physical one. As events of the sort they are, they can accordingly be dated, measured for intensity, and perhaps duration and rate, and the like.

On the other hand and more important here, sense impressions have properties which determine the specific act of sensing they individually are: an impression of red, perhaps, but not of blue; an impression perhaps of a red cube to the left of a blue triangle.

7. Sense impressions, since they are events, or happenings or processes rather than particular things, have properties in the way in which events or processes may have them but not as particulars do. An impression of red as an act of sensing and as a mental occurrence can scarcely be thought to be itself literally red. An impression of red is not a red impression.

8. Even though an impression of red is not itself red and is not even itself an act of awareness, it does nonetheless in a certain way "represent" or reflect a physical instance of that color quality. Individual acts of sensing belong to a range of acts which systematically mirror the patterns of exemplification and incompatibility of the sensory qualities and relations of physical things. Sense impressions have determinate properties which vary one with another as do also those which facing physical objects and happenings may exemplify. In this way, sense impressions exemplify features which reflect, but do not semantically represent, aspects of the physical objects which, ordinarily, initiate the sensory processes which terminate in their occurrence.

9. It is in this way that a sense impression is red, not literally, but "adverbially." The impression is the determinate sort of mental event which sighted, color-sensitive creatures tend to suffer when eyes open, from a middle distance, they face a red physical surface. But the very same sort of experience may of course occur as well in the absence of anything red in the physical environment. One cannot then validly infer from an experience of red that there is anything red which one experiences, for the sensing itself does not exemplify the color. Such an inference, "the sense-datum inference," is fallacious.

10. Nonetheless, there remains, Sellars holds, a related, valid, inference to be drawn from the occurrence of a sense impression. For in a sense which is not literal, there is, given an impression of red, something which is (in its own non-exemplificational way) red and of which we can be aware. This is the sense impression itself. Reasoning to an impression and to the way in which the impression is red is an inference to an explanation. It provides a characterization of the role of sensing in perceiving. Sellars calls this the "sense impression inference."4

11. It is because the sense impression in its (adverbial) way manifests (althought it does not exemplify) the quality sensed that an act of sensing, which is wholly nonconceptual, "guides" the judgmental, conceptual, element in an act of perceiving.

12. Sense impressions are part-causes of the conceptual acts involved in perception when these occur. An act of sensing is a necessary condition for the occurrence of an act of perception.

13. More interesting is the fact that a sense impression is also a part-determiner of the particular content of the conceptual acts which occur when one perceptually judges what may be the case. The conceptual acts of reference and ascription which occur when one perceptually judges what is the case are acts which have their own complex heritage. These conceptual acts are a function of many things: of the agent, his knowledge and maturation, his interest and attention, his physiological state and the state of his sensory apparatus. (Is the agent tired or distracted? Color-blind or tone-deaf?) They are a function of the physical setting and environment on the occasion of the occurrence of the act: of perspective, positioning, and of possible overriding events. But given these they are fundamentally a function of the perceiver's awareness of sense impressions. It is one's awareness of these and one's knowledge of the way their patterns of occurrence reflect standard causal properties of things, which underlie the sophisticated perceptual judgments we transact in a glance. Awareness of sense impressions without this knowledge about them is perceptual naivete. Knowledge of the relationships of sense impressions without awareness of them is not even perception.

Sense impressions "guide" the acts of conceptualization which are implicit in perceiving in the sense that these are for any given perceiver (of whatever particular perceptual sophistication) on any given occasion (of whatever physiological and environmental circumstance) directly a function of the particular impressions of sense of which he is aware.

14. This is not of course to say that all sensing leads to perception. As nonconceptual acts of consciousness, acts of sensing presumably may and do occur sometimes without any of the associated conceptual acts necessary for perception. There can be sensing without perceiving.

There is a further point as well. For, as nonconceptual mental acts, sense impressions occur not only without involving any awareness of anything further, but they may themselves occur without any knowledge of them. Without further assumptions concerning the nature and extent of self-knowledge, there is no reason to suppose that there may not be acts of sensing, as that has been so far characterized, of which we are not aware. And Sellars seems in fact to accept this consequence. Certainly, he denies that sense impressions are always apperceived, even urging the reasonableness of postulating that they are not apperceived at all.5 But if it is also reasonable to suppose, as we have, that sense impressions may occur without the occurrence of a related perception, then sense impressions may occur without being either themselves perceived or apperceived. And unless there is some other, so far implicit and unstated, mode of self-conscioussness, then it appears to follow that there may be sensings of which their agents are unaware. Thus, on the view, there may indeed be unfelt pains; and there may be impressions of color without any awarenesss of color.

So far, we have it that perception is a complex affair. It consists of non-conceptual items, sense impressions, and of conceptual ones. The conceptual items in the simplest cases include (but in ordinary cases outrun and presuppose rather than include) certain demonstrative acts of reference (which have been called "minimal perceptual takings") and they include certain conceptual acts of predication or attribution of varying degrees and kinds of complexity. We have summarized some of the important features of sense impressions, the nonconceptual element which is present in all perception. We turn now to the conceptual side of perception, with some comments first on the nature of minimal perceptual takings and on their relation to sense impressions.

2. Minimal Perceptual Takings.

It is, we already now know, Sellars's startling thesis that perceptions are not directly about what they seem to be about. They are not, as they seem to be, about the physical objects and happenings in the immediate, sensorily available, physical surroundings of their occurrences. Rather they are directly, demonstratively, about sense impressions, the very acts of sensing which they incorporate. What is startling about this is not the negative thesis, the claim that perception is not directly about physical things. That is familiar and common enough. We are used to versions of indirect realism as philosophical theories of perception. Nor is it unusual to be told that what we are directly aware of in perception is sense impressions. We have been told, often, that it is from knowledge of the occurrence of these that we infer the existence and disposition of those things in our physical environment which on the occasion affect the senses. What seems novel and striking in Sellars's account is his view of the way in which we are aware of sense impressions and given this awareness and the nonconceptual nature of sense impressions, of how perceptual knowledge is indeed knowledge of our surroundings. Sellars is not a classical indirect realist, as I understand them and him. For one thing, on his view, perceptual knowledge of our physical environment is not necessarily a matter of inference, either explicitly or implicitly. Perceiving occurs instead as though direct realism were true, as though we directly and non-inferentially perceived physical things. But we don't really do so, as science attests. But neither, then, do we "really" infer how things are given what we sense. The "fact" rather is that sense impressions are so intimately related to our acts of conceptualization, and they so finely and, for the most part accurately, reflect our surroundings that we successfully, for the most part, make our way around reality as though our sense impressions were direct awarenesses of qualities and relations of external things. The startling fact is, on Sellars's view, that if we use the word 'perceive' for that of which we are directly aware in a non-apperceptive way, then it is sense impressions we perceive and not physical things.

Literally, of course, this cannot be true. Literally, perception cannot consist of perceptions of sense impressions. It cannot, given our ordinary understanding of the term 'perception'. And it cannot, given our understanding of that term as Sellars strictly employs it. With respect to our ordinary understanding of the term consider, for instance, a visual impression. A visual impression, we have been told, is an act of sensing and not a thing. It is a mental occurrence, not a physical one. As such, a visual sense impression does not reflect light. Accordingly, it cannot itself literally be seen. The eye is sensitive to incoming, not internal, photo-stimulation. Awareness of visual impressions is thus not literally a case of visual perception, as we commonly conceive it.

Nor can we literally perceive sense impressions if we understand 'perception' in the technical sense we have taken from Sellars's general account. For on our characterization of that account, acts of perception are complex acts. They consist of constituent acts of sensing together with certain conceptual acts. Since acts of sensing are not themselves acts of awareness, a fortiori they cannot be acts of self-awareness. But if acts of sensing are (as Sellars postulates) the objects of acts of perception, then, given the nature of perception as characterized above, such perceptions of sense impressions must themselves consist of certain further acts of sensing. They must consist of ones appropriately related somehow to the sense impressions which are said to be perceived. Presumably, if these further acts of sensing should occur without our being aware of them, there would be no perception of the original sense impressions which are, by hypothesis, the "real" objects of perception. But if our awareness of these further acts of sensing is, as such, not apperceptive (as Sellars has stipulated that it cannot be,) then it must either be an act of perceptual awareness or some other form of direct awareness which has not yet been articulated by Sellars. But if we are perceptually aware of these further acts of sensing, then the cycle of explanation repeats itself. A vicious infinite regress is then implied. There exists an infinite regress on this alternative for the occurrence of any perception consists in part of the occurrence of a distinct perception of a constituent of the original perception. There is accordingly no perception without an infinite chain of perceptions. Such a regress is vicious for an arbitrary perception cannot occur without there being a completed infinite series of sensings with a perceptual awareness of each.

But if, confronting the regress, we hold that the awarenesses of the further acts of sensing, which are implicit in any given act of perception, are a special sort of awareness, then we need of course to be told precisely what sort of awareness this can be. What is it to be directly aware of a sense impression but to be neither perceptually nor apperceptively aware of it? And even if we do invoke some sort of special awareness to handle all this, we then surely need additional reasons for thinking that we perceive the original acts of sensing. Now that a special mode of awareness is available may it not be that we are aware of all sense impressions in this special way? Why should we suppose that just our sense impressions are perceived but that the further acts of sensing involved in this original perception are objects of some other special awareness?

Given the consequences of our ordinary understanding of the concept of perception, and of our "strict Sellarsian account" of it, it is natural to conclude that one cannot after all literally perceive acts of sensing. Pending a full characterization of the nature of -- and compelling motivation for belief in the occurrence of -- some special mode of awareness, it is not even clear that we do somehow have sense impressions directly before our minds as objects in acts of everyday perception.

All this may, I suppose, seem or be largely verbal. It results perhaps from an insufficiently sympathetic reading of Sellars. No doubt not too much should be made of the point. Perhaps the common idea of perception, as perception of external things, is here illicitly conflated with a certain philosophical and theoretical characterization of sensing and of its role in perception. Perception of external things, properly understood, is indirect. It is channeled through occurrences and conceptualizations of sense impressions. Sense impressions are acts of sensing of which we are in perception somehow directly aware. It is not to be expected, perhaps, that a term for what is in fact a general but indirect sort of awareness of external things should unambigously suffice to express the theoretical nature of the direct awareness postulated for nonconceptual acts of sensing.

Whatever we choose to call it, Sellars does have a bit to say about the direct awareness of sense impressions. This awareness is for one thing demonstrative. It is the sort of awareness which is present in acts of "minimal perceptual takings." It is an act of conceptual reference the verbal expression of which is or can be made articulate by the use of certain demonstrative definite descriptions. Such verbal expressions of minimal perceptual takings are, often, the grammatical subjects of the sentences which are used to express logically simple, subject-predicate, perceptual judgments.

A favorite Sellarsian example of a minimal perceptual taking, of that sort of reference which underlies simple perceptual judgments, is one a verbal expression of which might be this: this cool, smooth, transparent, pink cubical object. Such a reference is, perhaps, the subject of a perceptual judgment that what is before one is a pink ice cube.6

Some of the important features of minimal perceptual takings which are postulated by, or implicit in, Sellars's characterizations of such acts are these:

  1. Minimal perceptual takings are acts of reference.
  2. They are, thus, conceptual acts.
  3. They involve awareness of acts of sensing.
  4. Each has the logical force of a definite description. An occurrence of one thus carries, at least by implication, a certain existential claim together with a uniqueness claim, and a certain attributive structure.
  5. Each is an act of demonstrative reference.
  6. The attributive force explicit but unasserted in these acts of reference, involves a certain classificatory conceptualization. It involves as well a certain sensuous modification of the classificatory conceptualization.
  7. An occurrence of a minimal perceptual taking is partly caused by, and its content is partly determined by (some of) the sense impressions on that occasion of the agent of the perceptual taking.
  8. Minimal perceptual takings are not directly about what they often seem to be about. They directly refer to sense impressions. They seem, often, to be direct awarenesses of physical objects and happenings in the sensorily available environment.
  9. Minimal perceptual takings may, however, be indirectly about external things. They will be so when their direct objects, the nonconceptual acts of sensing to which they demonstratively refer, are causally related to the physical environment in an appropriate way.
  10. Minimal perceptual takings are complex acts of reference which are present in, or are presupposed by, all acts of perception.

Simple, propositional, subject-predicate perceptions accordingly have the following form: This S-K is P. In a verbal instance of this schema, 'S' will be replaced by a term for a sense quality, a proper sensible (like pink), 'K' by a classificatory common-sensible (like cube). In the schema, 'This' is understood as a syntactical variable tor any suitable demonstrative term. Depending on the predicate which replaces 'P', sentences of more or less conceptual complexity will result. These will range from ones suitable for the expression of relatively simple perceptual acts of detection, discrimination, and identification, to more complex ones expressing perceptual classifications and awarenesses of the natures, tendencies, or material stuff of what is before one. (Questions naturally arise as to how conceptually rich the values of 'P' can be. It is an interesting and difficult question to justify theoretically where the line is to be drawn between what is perceived on a given occasion and what is inferred. These interesting questions, however, are ignored here.)

What is of present interest here and cannot be ignored is the characterization of those acts of minimal perceptual takings which carry the references of our simple perceptual judgments. These are acts the schematic verbal form of which is: This S-K. and the nature of instances of which was summarized in the ten numbered items above. There are at least three very natural and very central sorts of questions to raise concerning minimal perceptual takings. The discussion of these questions will occupy our remaining space.

A natural first question is naive and straightforward enough: How minimal can minimal perceptual takings be? The fact is, given the characterization above of minimal perceptual takings and given their indicated form, that these acts of perceptual reference are conceptually surprisingly rich acts. These are acts which achieve their perceptual reference in certain conceptually qualified forms of direct, demonstrative awareness. These are acts of reference which internally consist in sensuously modified classificatory conceptualizations. Is it really true that all perceptions, that the most minimal and fleeting perceptual awarenesses, really involve references which are so conceptually complex?

Consider, by contrast, certain simple act of detection. Perhaps you hear something, over there, move. You may, in the experience, be perfectly aware that something has moved. But you needn't have seen, nor need you have been able by the sound alone to make out just what it was that moved. You need not be able with any precision to specify the location of it. Nonetheless you have judged something to be the case and your judgment consisted just in your perception of it. It seems clear that a perceptual act has occurred. It seems clear that it is an act which neither consists in nor presupposes some minimal perceptual taking, as these have been characterized. For where is the necessary structure of attributive and classificatory awareness? There need be nothing grudging or uncandid in the report, merely, that you heard something move, and that's all. The act of reference is unspecific as to subject and location. Movement has been detected, but there is not even discrimination of its source. Apparently, then, not all perceptions involve minimal perceptual takings, as these have been characterized; not all perceptual judgments involve references of the required conceptual complexity.

Consider, again, catching momentary trace in the air.7 The odor you detect may be as definite as one could wish. You could perfectly well identify, pick out, that particular kind of fragrance again. What you can't do, however in many cases like this, is specify or locate the source of the odor or classify its kind. In this instance, you can both detect the odor and discriminate it from others that waft your way. You distinguish it from the smell of newly cut grass, say. What you can't do is tell what it is you smell. You would recognize that kind of smell again, but cannot classify it. You could not even reidentify the same instance twice over, either in itself or by its source, as the same instance of the odor. It is doubtful, I believe, that perceptual references like these have after all the form, 'This S-K'.

There are thus in familiar perceptual occurrences like these grounds for skepticism concerning the character and ubiquity of Sellarsian minimal perceptual takings. But there are antecedent theoretical reasons for being suspicious of these anyway.

If minimal perceptual takings are indeed either literally present in, or are presupposed by, all acts of perception, then what can be the truth conditions of our ascriptions of such judgments to others? Surely, a necessary condition of the truth of a de dicto judgment of the form: Jones perceived that this S-K is P, is that Jones on the occasion perceived an S-K as such and knowingly. But surely for Jones to perceive something before him as an S-K is for Jones to see that what is before him is a K and that it, that very K, is S. This is to say that our original ascription of a perceptual act to Jones will be a true ascription only if a de dicto judgment of the form: Jones perceived that this is a K and it is F, is a true ascription of Jones, But this of course is something which is true to say or judge only if there are perceptions of the simple form: this is K.8

Perceptions of this form certainly are not perceptions whose perspicuous verbal representations literally include expressions for minimal perceptual takings. So, apparently if minimal perceptual takings are indeed either included in, or presupposed by, all perceptions, then for these perceptions such acts of reference can only be presupposed. But this, too, leads to an infinite regress or circle. Minimal perceptual takings imply the existence of perceptual acts of simple, unqualified, demonstrative reference. These, in turn, presuppose the existence of (other?) minimal perceptual takings. And even so, perceptual acts of simple, unqualified, demonstrative reference in any case remain as a kind of foundation of ordinary perceptual judgments. These remain not only as a necessary condition to an adequate semantics of Sellarsian minimal perceptual takings themselves, but quite as commonsense, and the occurrence of quite ordinary, simple perceptual acts of detection and discrimination, and (one reading of) traditional empiricism each suggest.

The misgivings now expressed about the characterization of minimal perceptual takings and earlier about sense impressions are heightened when one considers how, on Sellars's view, these mental occurrences are related one to the other in acts of perception. They are related, respectively, Sellars maintains, as awareness to object of awareness. This relation is however one of systematic misrepresentation. It is a relation between an act of demonstrative reference and its object of reference which is categorically mis-perceived in the act of demonstration itself.

There are two main elements at least to this characterization which evoke rather natural suspicions and motivate some pretty general, but natural, questions. The first concerns the element of systematic misrepresentation; the second, demonstrative reference to one's own mental states.

Minimal perceptual takings are references to acts of sensing. These references take, but do not assert, their objects to be external physical things or happenings. The implications of this are quite radical, but largely (for the most part and for most of us) unremarked and unsuspected. It is a consequence, for instance, that there really is nothing at all which is literally red, given the view. Sunsets, and clown's balloons, and my copy of Science and Metaphysics are not. For consider my perception of the cover of my copy of that book by Professor Sellars. The cover itself is that material cloud of particles or that set of physical energy states which causally initiates the sensory process which terminates in my sense impressions. This cover is not itself actually red. It is not, for that scientific reality is not literally, according to Sellars, colored at all. So, 'red' in 'red book' does not actually, but only through a kind of mental misdirection, apply to the book I see.

On the other hand, however, the actual object of my perceptual reference is not literally red either. For the actual referent is an act of sensing. This act is, in optimal circumstances, indeed "in-a-sense-red," as we have seen earlier. It is "red" in a sense to be explained by the certain systematic patterns of variations and invariances it shares, other things equal, with its initiating cause. But as a mental event or process it is not itself literally, and in the dictionary sense, red. The act does not itself exemplify redness. The adjective 'red' which I use to modify my reference to Sellars's book when I give verbal expression to that reference is not the right modifier with which adverbially to qualify correctly the mental events or processes of which we are actually, on the view, aware. There is, then, nothing red. There is nothing physically red and there is nothing mentally red. There is only our concept of that color and our tendency to depict things as colored. There is just the misdirected tendency to discriminate our individual acts of sensing as though they were instead commonly available physical things.

The sensory concepts we have all along used, the dictionary definitions of our terms for these concepts and for other terms whose correct usage is determined by these, all of these have no literal, correct application. It is not as though there might be on the view some exotic instance, a sense datum perhaps, which is held to exemplify the color I think I sense. There are, on the view, no proper sensibles at all. There are no literal instances of them, external or internal, of which I might have impressions.

On this view, I directly pick out my sense impressions of my copy of Science and Metaphysics in the act of taking the impressions to be what they are not, a nonmental physical object. Doing so, I may go on perceptually to judge that the object of my perceptual taking is somewhere out there before me; that it is some six feet from me on the table. Of course, the object of my reference is not, on the view, so located. That object is in my mind, being a mental act of sensing. So my reference is quite literally rnis-taken: my attribution is accordingly misapplied; and my judgment itself is false. But on the other hand, what I thought I perceived (the book) is as I thought I perceived it. The book indeed is on the table (or so the sum of all available sensory evidence combines to suggest). So if I nonsensuously judge that the book is on the table, then what I judge is true. Yet if I claim to see that very thing, if I perceptually judge that the book is there, my claim and my ostensible perception are false. The verbal expression of what I nonsensuously judge and what I ostensibly perceive may be the same. The concepts in terms of which I conceptualize what I do may, in the two cases, coincide. Still, I cannot actually have perceived what I may otherwise have actually and truly, nonsensuously judged.

Sellars is somewhat nonchalant about consequences like these. There are suggestions made of possible, future, conceptual revisions; of more accurate and scientifically more insightful applications. But it is not as though (pace Quine) this is like a case of replacing a commonsense 'if. . . then . . .' with a truth-value-gapless material conditional. It is not like this, for we have there an articulate theory of the material conditional. We know how to operate with the theory. We can imagine what it might be like to think in terms of it. But the fact is that we do not have an articulate theory of 'red' as an adverbial. We do not know what it would be to think in terms of a system of concepts which replace those for the proper and commonsensibles by modifiers of our states of sensing.

We know, for instance, how, in the "manifest image" of commonsense, demonstratives are used. We know, e.g., how to correct a reference made with a demonstrative definite description when that description misfires. We can distinguish failures of reference from an occurrence whose reference is vague, and from a felicitous one whose reference is nonetheless achieved by misdescription. But we have, by contrast, no articulate understanding of how minimal perceptual takings demonstratively refer to our own mental states. Is it possible to misfire with these demonstrations? And in the ways in which in the manifest image we seem to? What are the locational adverbial counter-parts which fix and distinguish the mental references we make? Can I, in some appropriate surrogate sense, mislocate my sense impressions and mistake one act of sensing for another?

There certainly may be, or there may come to be, articulate and insightful answers, theoretically justified, to questions like these for all we now know. But we don't now have such answers. Lacking them, Sellars's revisionary conceptual anticipations may well strike a skeptical reader as instead a way of making evident the consequences of what is best viewed as a reductio ad absurdum demonstration.

In the quotation which initiated this paper, Sellars asked whether it was necessary to introduce sense impressions as nonconceptual mental states whose postulated existence would suffice to explain those facts which any adequate theory of perception must explain. But is it really necessary to suppose that acts of sensing are not awarenesses of physical things; to suppose that all perception is direct awareness of acts of sensing? Sellars disparages what he thinks of as a contrary temptation to conflate minimal perceptual takings with the their objects, sense impressions. But in view of the consequences and implausibilities of the view he favors, why should we not suppose just that? Why should we not assume that, in the natural order, sense impressions are to perceptions quite as the tokens of the words we produce conventionally are, in the linguistic order, to the judgments we express through them? Evidently, the thoughts we so express are not (ordinarily) about the words by which we express them. Evidently, the material word-tokens we produce are not thereby deprived of their intentional character. Why, on an adequate theory of perception, should it of necessity be any different for sense impressions and for the perceptions which incorporate them?


NOTES

1. Wilfrid Sellars, Science and Metaphysics (London: Routledge & Kegan Paul, 1968), p. 16. Sellars's discussions of the themes with which this paper is concerned are extensive, subtle and profound. It would be difficult to overestimate their impact on contemporary academic philosophy in the United States. Since my understanding of these is no doubt callow, I fear that I may have misrepresented what Sellars holds or may have failed to notice crucial points altogether. In addition to his Science and Metaphysics, I rely here mainly on his "Berkeley and Descartes: Reflections on the Theory of Ideas." Studies in Perception, Peter K. Machamer and Robert G. Turnbull eds., (Columbus, OH: State University Press, 1978), pp. 259-311, and a longer manuscript version of "Giveness and Explanatory Coherence," portions of which originally appeared in The Journal of Philosophy, 70 (1973): 612-24.

2. The reader is asked to understand the attribution of views and doctrines to Sellars as carrying always a tacit rider, 'as I understand him'.

3. Roderick Chisholm, "The Theory of Appearing," Philosophical Analysis, Max Black, ed„ (Englewood Cliffs, NJ: Prentice-Hall, 1950), p. 107.

4. Science and Metaphysics, p. 17.

5. Ibid., p. 11.

6. See the longer manuscript version of "Giveness and Explanatory Coherence," p. 38.

7. I borrow this example from a paper of mine, "What Is a Perceptually Well-Defined Individual? Hintikka's Views on Perception," forthcoming in Profiles.

8. Sellars seems to agree. See: Studies in Perception, p. 279. But contrast this with the following pp. 280-85. Sellars seems to believe that detection and discrimination require criteria of identification.