Published in Hector-Neri Castaneda, ed., Action, Knowledge, and Reality: Critical Studies in Honor of Wilfrid Sellars (Indianapolis: The Bobbs-Merrill Company, 1975): pp. 109-127.

The Sensuous Content of Perception

ROMANE CLARK
Indiana University

A man sees a small, bouncy, red, rubber ball. There is some one thing he has seen, but how much seeing has he done? How many perceptual acts have occurred? Adequate philosophical theories of perception should, no doubt, provide schematic characterizations, instances of which would be, or yield, answers to questions like these.

In this paper I consider some aspects of Professor Sellars' views on perception, particularly with reference to the sensuous character of perception. I am only marginally concerned here with Sellars' broader philosophical themes concerning the naturalistic and materialistic bases of perception. I borrow, however, from these broader considerations his assumptions that judging (that something is the case) is best understood by analogy with overt speech, and that perceiving is a species of judging.

Some jargon will be useful in developing our themes. To avoid the "success" connotations of notions like 'see', and to generalize over the range of sense modalities, I shall often speak of (psychical) agents sensuously believing that something is the case or of agents being sensuously aware of something. The ascription of sensuous beliefs to an agent does not imply that what he believes is the case; nor does saying that an agent is sensuously aware of something imply that the object of his awareness exists. These idioms will be our perceptual analogues of 'thinks that' and 'thinks of'.

It is currently fashionable to construe the logic of the ascription of mental acts to psychical agents as a modal logic. [The fashion was created by Hintikka. See [3], [4].] If we combine current fashion with the assumptions borrowed above, it is plausible to think that the logic of the ascription of sensuous beliefs to agents is itself also thereby a species of modal logic. This is plausible, for, on those assumptions, perceiving is a species of judging. Moreover, if perceiving is a species of judging, and if judging is to be understood by analogy with saying, then it looks as though it should be fruitful to consider the analogy of seeing to saying.

Consider, now, the following three propositions:

  1. Perceptions are judgments.
  2. Sense impressions are constituents of perceptions.
  3. Sense impressions are not conceptual or cognitive items.

By 'judgments' we understand, not evaluation, but thoughts: those mental occurrences the verbal expressions of which are assertions. I borrow the term 'sense impression' from Sellars' writings. I do not know how closely it coincides with 'sensation' as that term has been used by other philosophers. Sense impressions must, however, at least include experiences of what Aristotle called the "proper sensibles." (How much more, if anything, the term 'sense impressions' covers remains open here.) The term covers. that is, at least experiences of those sense qualities awareness of instances of which would not occur if some single sense modality were destroyed. (If my eyes are put out, my capacity for experiencing instances of red (although not my capacity for remembering, or imaging, them) will be removed. Putting my eyes out will not similarly remove my capacity for experiencing instances of spatial relations. Presumably, when one is sensuously aware of some thing red, he often -- in normal circumstances at least -- has a sense impression of red. But whether one who is sensuously aware of a small, bouncy, rubber ball can have a sense impression of the size, nature, matter, or kind of what he is sensuously aware remains undetermined now.)

The truth of any two of (1)-(3) above may appear to imply the falsity of the remaining proposition. Evidently, qualifications concerning notions like 'constituent' will be required to turn these appearances into reality. We need in particular to know how, on a given theory of perception, sense impressions are thought to be related to sensuous beliefs. But in any case, this (putatively) inconsistent triad of propositions can be exploited to sharpen certain contrasts which exist among some philosophical theories of perception.

A naive realist, for example, might accept (1) and (2) but deny (3). (I note parenthetically that such a naive realist cannot be, in Sellars' sense, "an adequately critical direct realist" ([7]: 255; see also [5], [6]. He cannot be, for the latter, according to Sellars, affirms (3)). Everett Hall was such a naive (but inadequately critical direct) realist who accepted (1) and (2) but denied (3) ([2]).

One might accept (1) and (3) but deny (2). Coupled with a suitable gloss of 'constituent of', Sellars and Professor Geach seem to be philosophers who would take this line ([5], [6], [7); [1], esp. pp. 64ff, 122ff).

No doubt there are those who accept (2) and (3) but deny (1). (I have no one especially in mind.) Perhaps this is even the natural view, a view many or most of us shared before coming into contact with the contributions of philosophers like Brentano and Husserl, or Sellars or Hintikka to the Philosophy of Mind.

We ignore this last view here. We assume instead that (1) is true, that perceptions are judgments. We seek now grounds for accepting or rejecting (2) and for accepting or rejecting (3).

Evidently, (2), the proposition that sense impressions are constituents of perceptions, is in some sense true. In what sense it is true is not clear. For it is not clear in general what the constituents of judgments may be. We speak of mentally judging that something is the case and, in doing so, of thinking of an object and of mentally attributing something of it. We abstract and distinguish the judgment which results from these occurrent acts. But where in the judgment is the mental element of reference matching the act of thinking about something, and where is the mental element of ascription matching the act of attribution? And if, (1), perceptions are judgments, then perceivings, like thinkings, presumably are composite acts, and perceptions, like thoughts, presumably are logically complex entities. Where, then, in a perception is the referential element, and where in the perception is the ascriptive element of the perceptual judging?

It is a merit of Sellars' characterization of thought as a theoretical entity whose overt, external model is speech that verbally accommodating remarks about concepts as the constitutents of thoughts can be given some substance. The model is even more helpful, and more exact, for perception in particular than it is for thought in general. For a perception, like an assertion, is a logically complex entity the occurrent, related acts of which have material conditions and properties governing their manifestations. We distinguish, of course, assertions from assertings, the acts of making assertions. And we distinguish the sentence and the words uttered in the making of an assertion from the assertion and from the act itself. Further, sentences and words, we have learned, are themselves types, occurrences of which have their own material and occurrent properties. Some of these are, by convention, crucial features in determining what sentence or word a given token instance is. A given word instance, accordingly, is at once a material entity, an occurrent word instance, a word which, as used on an appropriate occasion, is, say, an ascription of a property. It is, qua token of a type, a symbol which occurs in the sentence used in making some assertion. The present point is that the material conditions and features of the word which are, by convention, fixed as the determinants of the physical occurrence determine it to be an instance of a conceptual element. The word, as a physical constituent of a sentence, is used to express a conceptual element of the sentence which is used to make a statement.

Something like this is true of sense impressions as mental occurrences which naturally, and not by convention, figure in the occurrence of sense perception. Note, at least, in the first place that the occurrence of sense impressions is a necessary condition for the occurrence of perceptions. It is of some importance to distinguish the reason for this being so from the fact, if it is one, that certain physiological states of the brain, say, are necessary conditions for the occurrences of certain judgments. It would be wrong to view perceptions as merely judgments plus some sense impressions. It is not as though a judgment exhausts the content of a perception with some correlated sense impressions serving as mere causal mechanisms, the material triggers for the occurrence of the judgment. For this would fail to distinguish perceptions from other judgments which may be materially induced by sensory stimulation, but which in no sense are perceptions, and which in no sense count as seeing what is judged. It may happen that I think or judge that the ball is red, that I am concurrently subject to the appropriate sensory stimulation, and yet be quite false that I see that the ball is red. I may not be attending to what is before me.

First-person, non-inferential knowledge of what I sensuously judge, like first-person, non-inferential knowledge of what I am thinking, is one of the motives for construing perception as a form of judgment. And occurrences of sense impressions are necessary conditions for occurrences of perceptions in this further, important sense: first-person, non-inferential knowledge of what I perceive is thereby also first-person, non-inferential knowledge of my current sensuous impressions. The fact is that a sensuous belief, the perception, say, that some object before me is red, requires a conscious awareness of the color of the object (although not, of course, necessarily a sense impression of red). Failing this, the belief is not a sensuous belief. It is not a perception of the object's color.

It is the occurrence of sense impressions relative to acts of sensuous belief which accounts for much of what is unique about perception as the species of judgment which it is. For sensuous beliefs differ from nonsensuous ones primarily in those elements of the perception which tie the act, the occurrent sensuous belief, to the occasion of its occurrence. We can detect, but cannot think of, something we know not what. We can think of, but cannot see, one who is absent. (We can, of course, have sensuous beliefs about one who is absent. We mistake, perhaps, something before us for him. What we cannot do, however, is have affirmative sensuous beliefs about him as being out of the range of the given sense modality.) Distance, obstructions, and perspective may prevent my seeing, but not believing, what is the case.

Sense impressions, then, are not merely necessary conditions of sensuous beliefs. They are constituents of them. Perhaps this is a truism. In any case, this fact is, in itself, of limited interest. Anyone who maintains (1) and (3) above, but denies (2), surely means to deny that sense impressions are themselves conceptual or cognitive constituents of sensuous beliefs. The considerations cited in reaching our truism suggest, I believe, that such a denial of (2) is wrong. But they do not force this conclusion. Everything depends, then, on now finding further considerations for determining the way in which sense impressions figure in sensuous judgments.

If we place this issue in the context of the general assumptions which initiated the paper, and which were borrowed from Sellars' writings, an interesting thesis rather naturally falls out. Since it is not a thesis which Sellars shares, and since it concerns the role of sense impressions in perceptions, it will be of some interest to develop it and to compare it with what Sellars actually maintains about sense impressions.

The initial assumptions, we recall, were that judging is best understood by analogy to speaking, and that perception is a species of judgment. Evidently, a certain strong, formal resemblance exists between those statements in which we, as scribes recording the occurrences of the events, ascribe certain assertions to individuals and to those statements in which we ascribe certain thoughts to them. Evidently, there is a strong, formal resemblance between assertions and judgments in their own rights. Sellars has written extensively concerning speech as the model for thought. We assume here, without review, the "Thought-Speech" analogy and Sellars' characterization of it ([5], [6], [7]).

The "Thought-Speech" analogy, plausible as it may seem, is, as described by Sellars, less persuasive when perception (which is, on present assumptions, a species of thought) is compared with speech. For one thing, there is the familiar fact that experiences, and so our sensuous beliefs, outrun our powers of description. This is not simply the "one-picture, a-thousand-words" theme. In principle, the situation is worse. I may, for example, have to show you what I cannot tell you. The fact is that colors, for instance, shade continuously from hue to hue but not so our words for colors. I can detect, discriminate, and even reidentify sounds, traces in the air, and perhaps peripheral movements, each being things the recognition of which I can only teach you by confrontation. David Kaplan somewhere put it like this, "If we cannot even say it with words, but have to paint or sing it, we cannot believe it with words." It might once have seemed plausible to insist that I think in English, say. But it is not at all plausible to insist that I perceive in English. I do not. It follows that if sensuous believing is a species of thinking, the "Thought-Speech" analogy is, in one version, misleading. The model of thought as an internal version of overt speech misses an important feature of perception. And the feature which it misses is directly related to the occurrence and nature of the sense impressions which occur in our sensuous beliefs.

It is of some importance to note that the present failure of the analogy is on the side of the model, of overt saying. The point is not that the palpable model entities have features which the alleged, theoretical, hidden entities being modeled do not. That is familiar enough and acceptable. Speech after all is linear and used for communication, features which no one requires of thought. The failure lies the other way around. This is to say that the analogy misses instead something essential to what is modeled, something essential to the nature of sensuous belief. My complaint is not that the "Thought-Speech" analogy reverses our priorities, explaining thought in terms of speech, rather than the other way around. (I think it does, in fact, do this but that is another issue My complaint is rather that the analogy fails to do justice to what made it seem plausible and helpful in the first place.

If one stops to think about it, it is quite amazing that Sellars offers a model of thought based upon overt speech acts and yet offers a quite distinct and unrelated model of the occurrence of sense impressions. Yet perceptions are a species of thought, and sense impressions are (however their role may be characterized) intrinsic to perception. Somehow, one would have thought, the model of sense impressions has got to get linked with that of thoughts. Otherwise, our theories of thought -- and so, too, our theories of perceptions as well, as a special species of thinking -- should be theories quite unrelated to the occurrences of sense impressions. Or, turned around, the present point runs counter to Sellars' separation of the models of thought and sense impressions in a quite direct way: since a theory of thought must include, that of perception, and since occurrences of sense impressions are necessary conditions for the occurrence of acts of perception which they constitute, there can be no adequate, general model of thought which does not thereby model the role of sense impressions in the making of sensuous judgments as well.

Sellars' model of thought we know to be overt speech. Sellars' model of sense impressions is sketched like this:

Analysis reveals a second way in which the sense of "impression of a red triangle" is related to the sense of "red and triangular physical object." The first has already been characterized by relating "S has an impression of a red triangle" to "S is in that state [brought about in normal circumstances by the influence of red and triangular physical objects on the eyes] etc." The second consists in the fact that visual impressions of red triangles are conceived as items which are analogous in certain respects to physical objects which are red and triangular on the facing side . . .

(a) Impressions of red, blue, yellow, etc. triangles are implied to resemble-and-differ in a way which is formally analogous to that in which physical objects which are triangular and (red or blue or yellow, etc.) on the facing side resemble-and-differ; and similarly mutatis mutandis in the case of other shapes.

(b) Impressions of red triangles, circles, squares, etc. are implied to resemble-and-differ in a way which is formally analogous to that in which physical objects which are red and (triangular or circular or square, etc.) on the facing side resemble-and-differ; and similarly mutatis mutandis in the case of other colors. ([7]: 258-59.)

Sellars' model of sense impressions (as states of a psychical agent) is, thus, that of a bunch of physical objects with certain properties. The structural relations which exist among the manifestations of these physical attributes model those which, on his theory of mind, obtain among certain states of the agent.

Evidently, then, Sellars' models of thought and of sense impressions are radically unrelated: they are literally as diverse as chips of wood and guttural mouthings. Unrelated as they are, not both can be satisfactory models for the full range of thought, including as that does sensuous judgments and so also sense impressions. One of the models, at least, must then go. The question is whether either model is salvageable and throws illumination upon our original inquiry. That inquiry was concerned with the question: What is the relationship of the sensuous content of our acts of perception to the perceptual acts of which they are constituents? What can be an adequate model of thought which will reflect this relationship?

What made the "Thought-Speech" analogy plausible in the first place resided in some very general, very pervasive features common to each. Both thought and speech are intentional; each has truth-values; the ascription of the occurrence of an act of either kind to an agent exhibits formal similarities to the ascription of the other kind to him; each is the subject of immediate, reflexive awareness; and so on. None of these similarities, of course, requires, or even suggests, that thought be viewed as internalized speech. A view like that is a further consequence of some given philosophical theory of thought. The analogy supports as well (and also as little) the opposite view: the view that speech is mere externalized thought. Neither of these further philosophical stances is relevant here. What is relevant here is not the priority of thought to speech, or of speech to thought. What is important is the fact that these analogous features do exist. The fact that they do exist ensures that, so far forth, then, speaking is a plausible, overt model for the covert mental occurrences which we call thinking. The intentionality of thought, for instance, is analogous to the intentionality of statement. And this is a fact of great importance.

The further point is that these very general and very pervasive features are shared by perceptions as well. Sensuous beliefs, like non-sensuous judgments and like overt sayings, are intentional, have truth-values, and are the objects of reflexive self-knowledge. They exhibit similar, common, formal features as well. It is worth remarking that in these important respects, the disanalogy we noticed earlier between seeing and saying, the fact that sensuous experiences outrun our capacities for verbalization, no longer counts against the model. For the model is no longer a model of thought as internalized speech, but a model of perception as sharing the intentionality and general formal properties of the ascription of assertions to speaking agents. Given this account, moreover, a perfectly natural model of sense impressions is forthcoming. It falls out as a by-product of the model of a full theory of thought, a theory which includes, of course, sensuous thinking.

The model of sense impressions, implicit in the model of thoughts, is simply this: as predicates are to declarative sentences, as ascriptions are to assertions, as concepts-are to judgments, as ascribing is to stating, so, too, sense impressions are to perceptions. Predicative ascriptions of qualities to objects in ordinary assertions of fact are the overt analogues, the models, of sense impressions as constituents of sensuous judgments. The using of conventional entities to characterize objects of reference is analogous to the having of natural entities -- sense impressions --- in seeing how an object is. In the natural, non-conventional "language" of perception, sense impressions are ascriptions of sensuous qualities to the objects of perception. Sense impressions are the predicate "words" of perceptual, mental "assertions." The model goes through unusually well. For conventional words reveal and highlight features necessary to our characterizations of sense impressions. As we remarked earlier, conventional words are, materially, physical entities the production of which is the conventional mechanism by which we as agents perform certain functions necessary to the act of assertion. Equally, the having of sense impressions is the natural material base for the occurrence of acts of perception. The occurrences of sense impressions in acts of perception are the vehicles for the ascription of qualities to what is before one as the use of predicate words in the making of assertions is our way of describing an object of reference. As predicate words, the use of which is essential to assertions, are conceptual elements of assertions, so too, sense impressions, the occurrence of which is essential to perceptions, are conceptual elements of perceptions. Thus, the occurrence of a sense impression is the predication of a sense quality to an object, and so it is in this way that sense impressions are cognitive or conceptual elements of sensuous beliefs. That sense impressions are constituents of sensuous judgments in just this manner amounts to accepting the truth of (1) and (2) but the denial of (3) in the original trichotomy of propositions which initiated our discussion. It is, of course, accordingly, a view which is incompatible with that familiar from Sellars' writings. Sellars, we know, affirms (3).

The analogy between conventional quality words and sense impressions is, I believe, strikingly apt. Fundamentally, there is, of course, the familiar ordering of determinates under determinables in patterns of incompatibility. The record of these relationships, fixed in the correct usage of color words, is the conventional reflection of those quasi-analytical truths of perception that, e.g., nothing can be (at once) both red and green (all over). The sense impressions, falling within a given sense modality, are the primitive base in experience of such structures. This has implications within the theory, as we shall see later, in determining the identity of a single, basic, sensuous judgment.

But other similarities between sense impressions and quality predicates go through as well. There are, for instance, in a quite direct and not altogether trivial sense, sensuous homonyms as there are verbal homonyms. For just as two words may be identical in sound but differ in meaning, the contexts of their occurrence making, perhaps, the intended ascription clear, so also two sense impressions may, in their occurrent sensuous content, be identical while the contexts of their occurrence may make it quite evident that different qualities are ascribed by these occurrences to what is before one. A sense impression of an elliptical shape, say, in a given context, is, we know, a necessary constituent of the perception that the object seen from a certain perspective is round. This is, of course, something to be learned, case by case. Children and foreigners come to distinguish not merely words of different sound, but words which have the same sound but different meanings. So, too, we all come to distinguish the import of sense impressions, qualitatively indistinguishable, by the contexts of their occurrences. This is possible because we distinguish the occurrent properties of sense impressions and their semantical functions. We do this quite as we distinguish the occurrent properties of quality words and their semantical functions. This distinction itself is further evidence of the analogy of sense impressions to predicate words and the appropriateness of the model.

Words, of course, can be attended to in their own right. They can occur in isolation; they can be given voice, perhaps, solely for their amusing or pleasant euphony. They can be abstracted from their occurrences in the contexts of assertion. We can reflect upon them as items with their own occurrent properties. But so too are sense impressions items which can occur in isolation, which can be savored for their own characters, and which can be reflected upon in abstraction from their occurrences in the contexts of perceptual judgments. Of course, the capacity to recognize words as occurrent entities which are qualitatively the same is a necessary, though not sufficient, condition to perceiving some of them as homonyms. Equally, it is the capacity to recognize sense impressions as the same in their qualitative content which is a necessary condition for the acquisition of the distinction between how things appear and how things are, and for distinguishing occurrences of sense impressions which, although they are the same in their sensuous natures, nonetheless function in the different contexts of their occurrences as different ascriptions.

Semantically, predicates are sometimes formally viewed as those linguistic elements which are assigned functions. These functions take individuals as arguments into truth-values or perhaps facts. The homonyms of a natural language, then, are words which, though identical in sound, are, in the formal sense, assigned different functions. Homonyms differ in meaning at least in the precise and limited sense that, for certain arguments, they determine opposite truth-values. Homonyms are accidentally the same in certain occurrent properties. They are however quite different words with quite different meanings. The determination of which word an homonymous occurrence expresses is largely a matter of sensitivity to context. In practice, this involves discrimination both of its syntactic function in the larger linguistic context of its occurrence and of the pragmatic circumstances of the occasion of its utterance as well.

Something quite like this is true, I believe, of sense impressions as well. Viewed formally, we can think of these, too, as functions taking individuals into truth-values or facts. A pair of different sense impressions belonging to the same sense modality will, in contexts of occurrence which are otherwise the same, yield sensuous judgments which cannot both be true. And the same sense impression in different contexts of occurrence may yield sensuous judgments of opposite truth-values. Such impressions are, then, literally different sensuous ascriptions of what is before one. The determination of what sensuous ascription a given occurrent impression embodies involves sensitivity to the natural context of its occurrence and is a function of the state and knowledge of the observer as well. (These similarities of words and sense impressions are, of course, quite distinct from the further fact that a sense impression, of red, say, no more itself exemplifies the color it ascribes than does an arbitrary inscription of the word 'red'. This last one is, it happens, black.)

So far I have suggested that, contrary to Sellars' view of things, we accept propositions (1) and (2) and deny (3). I have suggested that perceptions are a species of thought, that sense impressions are indeed constituents of perceptions, and that sense impressions in fact are cognitive constituents of them. I have argued that, as constituents of perceptions, sense impressions play a role in natural experience analogous to the role conventionally performed by certain predicates in acts of assertion. (We have noted that, of course, it is predicates which occur in statements (in a theoretical sense which Sellars has been concerned to explicate in a certain manner). Predicate words, by contrast, are English or French words, say, and occur in sentences.) It is in this way, analogous to predicate words, that sense impressions are properly viewed as conceptual entities rather than mere causal mechanisms. This is the view of Everett Hall's naive realism ([2]). Hall, like Brentano, thought that all psychical occurrences are intentional and so, he believed, they are all conceptual occurrences. Hall, it is clear, would have felt quite at home adapting Sellars' "Thought-Speech" analogy to his own purposes. He did himself explicitly invoke the model of predication for the explication of sense impressions. Sellars has directly commented upon Hall's "intentional realism" ([6]). We shall turn to his criticisms and to certain hard cases for Hall's view in a moment. We pause now for a few brief, and so unfair, last comments on Sellars' own characterization of sense impressions and their material model. These comments are based upon the passages quoted from him above.

Granted that our visual sense impressions of colored triangles resemble and differ from one another in a manner formally analogous to the way in which triangles of various colors on the facing side do, what are we to make of the fact? What insight does knowing these things about physical objects convey about sense impressions? (Contrast this with viewing the occurrences of sense impressions in perceptions by analogy with that of predicates in assertions, and all that that entails.) The gist of Sellars' account lies in the existence of a one-to-one relation between a psychical agent's states and the manifestations of certain material qualities of physical things. Perhaps a central difficulty with the view can be most briefly put by noting the science fiction it tempts us to indulge in. For surely there are, or may be, indefinite patterns of structures on each side of this one-to-one correlation. The facing objects in their complex physical states may well exhibit all sorts of patterns of physical properties invariant with certain states of observers. And on the other side, we can imagine all sorts of brain states invariant with patterns of physical stimulatory bombardments. Indeed, if we are allowed (on pain of otherwise begging questions) to be liberal enough with the concept of an observer, the numbers of scattered or submerged or cross-linked physiological states that may exist in just the required patterns of occurrence are probably scarcely more limited than our imaginations. What this misplaced scientism shows about sense impressions is difficult to tell. But in any case the model does not illuminate what is special and interesting about sense impressions as mental phenomenon: the fact that, and the way in which, the havings of sense impressions are awarenesses of the sensible qualities of material things, or, more accurately, of the way in which they are impressions of sensibly qualified material things. What the model does presuppose, of course, is the causal link between environment and experience which ties the occurrent impression, in the standard case, to one's surroundings. That fact, however, was never in question. The determination of the causal mechanisms was never the philosopher's task.

I revert, instead then, to our earlier model, to the view that sense impressions function semantically like certain predicates, the quality adjectives of natural languages. Pointed questions like the following naturally arise. If sense impressions function in perceptions, semantically, as do predicates in ordinary assertions of fact, what then are their arguments? What, literally, are the occurrent references the application of certain sense impressions to which yields on that occasion a specific perception? Moreover, the set of perceptions, after all, consists not merely of ascriptions of qualities to the objects before one. It comprises as well recognitions and classifications of what is before one, of what something is made, or what a thing may do or be doing. Each of these is something that we on occasion sensuously judge. Each is something that we can know or believe by perception. How, then, are sense impressions related to these other sensuous beliefs? Finally, we need to know and say something about the nature of perceptions as judgments. What is their logic? How many judgments.does a given perception embody? How does one tell? How should direct-object constructions, the sensuous awareness of objects, be construed? In what does the distinctness and identity of a sensuous judgment consist? These and other related questions are questions of what, on a given theory, sense impressions and sensuous judgments come to. They are accordingly questions upon which the present account of sense impressions must hazard answers, if it is to be at all a plausible candidate as a theory of the sensuous element in judgment. Some of these are questions which Sellars raised in his examination of Hall's "intentional realism" ([6]). The remainder of this paper is devoted to some comments on these questions.

Earlier, we employed a misleading idiom, speaking of the sense impression of red. The underlining is a symptom of a theoretical embarrassment. Our sense impressions are not of qualities but of qualified objects. I see, not colors, but (in the standard case) colored objects; I hear, not sounds, but sounding objects: I feel, not pains, but portions of my anatomy that hurt. My visual sense impression is an ascription of a color, say, to that which is visually before me, not to the impression itself. But what, for example, is, in the simplest cases, before me? To what is the color ascribed? By what natural forms does singular reference manifest itself in sensuous judgments?

No doubt in common cases the object of awareness to which the sensuous ascrption is applied is some instance of a natural kind. No doubt in common cases natural reference to these comes via a sensuous recognition or identification of the instance as a certain case of the given kind or via a reidentification of the same object, But these common cases, familiar as the may be, are theoretically too complex for the level of analysis to which we aspire here. We do not have sense impressions of ball or rubber, and so we do not have epistemologically primitive singular references to balls or to things of a given material nature, as being such. Sense impressions are primary occurrences in the acquisition of knowledge. But sense impressions have secondary importance in explanatory power in the corpus of our sensuous beliefs. In general, the sensuous awareness that what is before me is a car ranks higher in the intellectual scheme of things than seeing something, I know not what, to be green. Evidently, then, a theory of sense impressions requires, as a theoretical characterization of the acquisition of knowledge, the ascription of sense qualities to entities prior to the later classifications which they may help to make available of those entities. We may think of perceptions of these minimal sorts as "basic perceptions." These are basic in two senses. First, they are ascriptions of simple sensuous qualities, the proper sensibles. Second, they have no internal, logical complexity. Each is simply a (single) qualitative ascription to a (single) object of reference. What are the vehicles of singular reference in basic perceptions? Hall put the point paradoxically: ". . . I have perception which are descriptive or predicative throughout . . . And Sellars replied with puzzlement: ". . . how [can] a pure perception . . . be a sentence, and yet be 'predicative throughout'. . . . Must not pure perceptions contain expressions referring to an object in order to be able to characterize an object?" ([6]: 110.)

Sellars is right. Perceptions, if judgments, must contain referring elements. A fortiori, basic perceptions must. There are two comments to be made in explaining how this requirement, which must be fulfilled, is fulfilled, although basic perceptions nonetheless consist solely of sensuous ascriptions. The first comment is a mere palliative. Basic perceptions, we note, are required by theory, not necessarily encountered in practice. It may never be that an articulate user of language has or recalls, basic perceptions at the point at which he has mastered language in a serviceable way. These are idealizations postulated to account for the empirical basis of the acquisition of the more complex, but standard, forms of sensuous belief.

Idealizations or not, the question requires an answer. It is true in a sense that basic perceptions are ascriptive throughout. Their sensuous content resides in the sense impression of which they consist. But they have a reference. And if the reference which they have is not itself carried by some articulate conceptualization of individualized, embodied, sense qualities, then it must be carried demonstratively. And it is. The sheer occurrence of an impression of a proper sensible, any at all, is in fact a putative ascription. We may later come to learn when to withhold such predications. But in basic perceptions, the occurrence of a sense impression in the context of its occurrence provides itself the demonstrative element of the sensuous judgment. Sense impressions are, after all, impressions of sensibly qualified objects putatively before one within the range of the given sense modality. It is the sensuous content the occurrent impression which in the context of its occurrence determines the ascriptive element of the basic perception. It is the material occurrence of the impression in the given context which provides the demonstrative reference to that experience.

We can exploit Sellars' own writings to make our present point. In "Naming and Saying," he considers a perspicuous language in which predication is carried not by the juxtaposition of referring and predicate expressions but by writing names so that they manifest the qualities or relations to be ascribed their referents. Thus, writing the name "John" in red might be the vehicle to state what we should put in words by writing "John is red." In keeping with the "Thought-Speech" analogy, Sellars' perspicuous language provides the perfect linguistic vehicle to model basic perceptions. Basic perceptions are demonstrative occurrences which ascribe the content of the occurring sense impression to what is before one. The model of the basic perception that this before me is red is, in the perspicuous language, accordingly an occurrence of the demonstrative "This" written in red. Or, vocally, an occurrence of the vocable "This" uttered shrilly would be the vehicle by which we state that this is shrill. Just as the occurrence of a demonstrative is, in appropriate contexts, a reference to something in the surroundings, so too the occurrence of a sense impression is a reference to something before one of which he is aware. And just as the vehicle of reference in the perspicuous language manifests what is ascribed to the object of reference, so too the content of the sense impression is ascribed to that to which its occurrence makes demonstrative reference.

Semantically, sense impressions are functions taking the contexts of their own occurrences as argument into truth-values or facts. They yield (perceptual) truths if the demonstrative references of their occurrences have the sensuous characters which their contents ascribe. The occurrence of a sense impression is an ascription the verbalization of which would run schematically thus: "This, sensuously before me, is thus and so," (All psychical phenomena, Brentano thought have objects, and the occurrences of sense impressions are, after all, mental occurrences.)

This view has further implications for a theory of perception. Basic perceptions are now to be identified via the sense impressions which occur in them and the contexts of their occurrences. This means, for instance, that the following perceptions -- this before me is red and round; this before me is red and sticky; this before me is red and that before me is green -- are not individual perceptions of conjunctions of sense attributes of things. Common, complex perceptions of what is the case in standard acts of awareness not only are not single basic perceptions, but on the present account, involving, as they typically do, awarenesses of the kinds and natures of things, they are not basic perceptions at all. They are, on the present account, clusters and groups of what, upon analysis, presuppose individual sensuous ascriptions of simpler kinds. The psychical agent, then, who sees that the object before him is a small, bouncy, red, rubber ball, has sensuously judged that something before him is red and has judged a whole lot more besides. (How much more, and how, the difficult questions which really interest us, are not questions the themes for answers to which I can provide here.) It is each basic perception, each ascription of a sense impression, coupled with the contextual reference of its occurrence, which counts as a single, distinguishable, minimal, perceptual act.

Sellars, by contrast, commenting upon clusters of ascriptions, puts it thus:

The point at which I am driving is, in Kantian terms, that the perception of a manifold must not be confused with a manifold of perceptions. Thus, if we leave 'and' out of the language of pure perceptions, one could not pure-perceive an object to exemplify a number of properties all together. . . . Thus it could be argued that conjunctive perception involves a "conjunction introduction move" from the component perceptions. The important thing would be that the inner Ianguage of perception proper contained the logical apparatus for making this move, so that the move could occur in it. ([6]: 112-13.)

But, surely, Sellars' Kantian introduction of the logician's categories into the forms of the manifold of sense cannot in general, at least, be right. It is not right, surely, for basic perceptions. A psychical agent, assume, sensuously believes on a given occasion that something before, him is φ. Does it indeed then follow that the agent believes, on that occasion that something before him is both φ and ψ? Since these are basic perceptions, the primitive experiential base of awareness by assumption on the theory, these perceptions are, at least sometimes, ascriptions of proper sensibles to what is before the agent. And if this is so, then, since one does not on the same occasion simultaneously have multiple sense impressions in the same sense modality of the same object, our agent's ascriptions must in fact be cross-modal ascriptions. (One, perhaps, is visual; the other perhaps tactual. Thus, the agent may know (by sight, say) that what is (visually) before him is red. And, on the same occasion the agent may know that what is (tactually) before him is sticky.) But surely there is no-contradiction, however broad our categories of rationality, in supposing that the agent does not know or believe on this occasion that what is red is what is sticky. If so, he need not believe that some one thing before him is both. Evidently, "conjunction introduction" is not part of the logic of "inner sense," if this is the logic of basic perceptions.

Unlike Sellars, I do not think of the logical connectives as part of perceptual judgments. Certainly not on the primitive level of basic perceptions. There are, in fact, some logical gadgetry and a philosophical stance which suggest viewing matters quite oppositely. We think instead of Gentzenian reductions, of backward reasonings; perhaps these are Peircean abductions, rather than the introduction into sense of Kantian categories. In any case, given that an agent does sensuously believe that something both φ's and ψ's (where perhaps these are the ascriptions of sense impressions), we may indeed posit both that the agent sensuously believes that this is φ and sensuously believes as well that this is ψ. As in all good theorizing, we assume in such a case the simplest sufficient conditions which will account for the fact to be explained. If the agent believes both things, then he must believe each and be in command as well of the concept of conjunction. But, note, this is not to be in command of a perceptual conjunction, whatever that may be.

It is, I believe, part of the logic of perception that the ascription of any perception to an agent entails that there exists a sense impression which the agent experiences. It is, further, part of the logic of perception, if the present account is right, that the ascription of the perception of a conjunction to a psychical agent implies a conjunction of such ascriptions to him, indexed to the same occasion. But the converse, we have argued, is not true. This is of some interest, for it suggests a seam at which the logic of sensuous judgment gets stitched to the total corpus of an agent's beliefs. Thus, while it does not follow from the fact that a psychical agent sensuously judges on an occasion that P, and also sensuously judges on that occasion that Q, that the agent therefore on the given occasion sensuously judges that both, still it does follow that he must (virtually) judge (albeit non-sensuously) on that occasion that both P and Q. For what one sensuously believes, one believes. And an agent who believes each of two things is, if rational, an agent who virtually accepts their conjunction. This is nice, for it suggests that we can keep a primitive experiential base for perception unsullied by the intrusion of Kantian categories of logic. On the other hand, the introduction of logical connections among beliefs and so, with it, the further possibility of relating certain complex and non-basic perceptions, e.g. of kinds, to basic sense impressions remains quite available.

We cannot further explore here the formal features of Hall's "intentional realism." [Some brief remarks of mine concerning quantification into epistemic ascriptions have been an object of criticism by Robert C. Sleigh, [8]. I hope to respond to these in detail soon.] It is clear how some of the questions which remain dangling must be treated by the theory. Direct-object constructions, for instance, give way on this view to that-constructions. One hears, not sounds, but sounding objects; in being sensuously aware of an object, one sensuously judges something of how it manifestly is. But however these further matters go, the main point which remains is this: an adequate theory of perception can characterize sense impressions as cognitive elements in perception without, thereby, incorporating the full, formal powers of conventional Ianguage into our primitive experiences. [I am indebted to Ausonio Marras for helpful comments and criticisms of a draft of this paper. There has been neither space nor time to make more than minimal corrections in the wording of the text; more substantial alterations and amplifications in the light of his comments must be postponed.]


REFERENCES

[1] Peter Geach, Mental Acts (London: Routledge & Kegan Paul, 1957).

[2] Everett Hall, Our Knowledge of Fact and Value (Chapel Hill: University North Carolina Press, 1961).

[3] Jaakko Hintikka, Knowledge and Belief (Ithaca, N. Y.: Cornell University Press, 1962).

[4] __________, "On the Logic of Perception," in Norman S. Care and Robert H. Grimm, eds., Perception and Personal Identity (Cleveland: The Press of Case Western Reserve University, 1969).

[5] Wilfrid Sellars, "Empiricism and the Philosophy of Mind," in H. Feigl an M. Scriven, eds., Minnesota Studies in the Philosophy of Science Vol. I (Minneapolis: University of Minnesota Press, 1956).

[6] __________, "The Intentional Realism of Everett Hall," Southern Journal of Philosophy 4 (1966): 103-15.

[7] __________, "Phenomenalism," in Hector-Neri Castaneda, ed., Intentionality, Minds, and Perception (Detroit: Wayne State University Press, 1967)

[8] Robert C. Sleigh, "Restricted Range in Epistemic Logic," Journal of Philosophy 69 (1972): 67-77.